research article numerical modeling of fractional-order biological...
TRANSCRIPT
Hindawi Publishing CorporationAbstract and Applied AnalysisVolume 2013 Article ID 816803 11 pageshttpdxdoiorg1011552013816803
Research ArticleNumerical Modeling of Fractional-Order Biological Systems
Fathalla A Rihan12
1 Department of Mathematical Sciences College of Science UAE University Al Ain 15551 UAE2Department of Mathematics Faculty of Science Helwan University Cairo 11795 Egypt
Correspondence should be addressed to Fathalla A Rihan frihanuaeuacae
Received 17 May 2013 Accepted 23 June 2013
Academic Editor Ali H Bhrawy
Copyright copy 2013 Fathalla A Rihan This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
We provide a class of fractional-order differential models of biological systems with memory such as dynamics of tumor-immunesystem and dynamics of HIV infection of CD4+ T cells Stability and nonstability conditions for disease-free equilibrium andpositive equilibria are obtained in terms of a threshold parameterR
0(minimum infection parameter) for each model We provide
unconditionally stable method using the Caputo fractional derivative of order 120572 and implicit Eulerrsquos approximation to find anumerical solution of the resulting systems The numerical simulations confirm the advantages of the numerical technique andusing fractional-order differential models in biological systems over the differential equations with integer order The results maygive insight to infectious disease specialists
1 Introduction
Mathematical models using ordinary differential equationswith integer order have been proven valuable in understand-ing the dynamics of biological systems However the behav-ior ofmost biological systems hasmemory or aftereffectsThemodelling of these systems by fractional-order differentialequations has more advantages than classical integer-ordermathematical modeling in which such effects are neglectedAccordingly the subject of fractional calculus (ie calculusof integral and derivatives of arbitrary order) has gainedpopularity and importance mainly due to its demonstratedapplications in numerous diverse and widespread fields ofscience and engineering For example fractional calculus hasbeen successfully applied to system biology [1ndash5] physics[6ndash9] chemistry and biochemistry [10] hydrology [11 12]medicine [13 14] and finance [15] In some situations thefractional-order differential equations (FODEs) models seemmore consistent with the real phenomena than the integer-order models This is due to the fact that fractional deriva-tives and integrals enable the description of the memoryand hereditary properties inherent in various materials andprocesses Hence there is a growing need to study and use thefractional-order differential and integral equations
Fractional-order differential equations are naturallyrelated to systems with memory which exists in most
biological systems Also they are closely related to fractals[16] which are abundant in biological systems It has beendeduced in [3] that the membranes of cells of biologicalorganism have fractional-order electrical conductance andthen are classified in groups of noninteger-order modelsFractional derivatives embody essential features of cellrheological behavior and have enjoyed greatest success inthe field of rheology [17] In this paper we propose systemsof FODEs for modeling the interactions of tumor-immunesystem (see Section 2) and HIV infection of CD4+ T cellswith immune system (see Section 3)
However analytical and closed solutions of these types offractional equations cannot generally be obtained As a con-sequence approximate and numerical techniques are playingimportant role in identifying the solution behavior of suchfractional equations and exploring their applications [18ndash21]The Adomian decomposition method [22 23] extrapolationmethod [24] multistep method [25] monotone iterativetechnique [26] and predictor-corrector approach [18 27]are proposed to provide numerical solutions for linear andnonlinear FODEs The monograph of [28] investigates theinterconnection between fractional differential equations andclassical differential equations of integer-order derivatives Inthis paper we provide a numerical approach for the resultingsystem using implicit Eulerrsquos scheme (see Section 4) We alsostudy the stability and convergence of the proposed method
2 Abstract and Applied Analysis
We first give the definition of fractional-order integrationand fractional-order differentiation [29 30]There are severalapproaches to the generalization of the notion of differenti-ation to fractional orders for example Riemann-LiouvilleCaputo and Generalized Functions approaches Let 1198711 =
1198711[119886 119887] be the class of Lebesgue integrable functions on [119886 119887]
119886 lt 119887 lt infin
Definition 1 The fractional integral (or the Riemann-Liouville integral) of order 120573 isin R+ of the function 119891(119905) 119905 gt 0
(119891 R+ rarr R) is defined by
119868120573
119886119891 (119905) =
1
Γ (120573)
int
119905
119886
(119905 minus 119904)120573minus1
119891 (119904) 119889119904 119905 gt 0 (1)
The fractional derivative of order 120572 isin (119899 minus 1 119899) of 119891(119905) isdefined by two (nonequivalent) ways
(i) Riemann-Liouville fractional derivative take frac-tional integral of order (119899 minus 120572) and then take 119899thderivative as follows
119863120572
lowast119891 (119905) = 119863
119899
lowast119868119899minus120572
119886119891 (119905) 119863
119899
lowast=
119889119899
119889119905119899 119899 = 1 2
(2)
(ii) Caputo-fractional derivative take 119899th derivative andthen take a fractional integral of order (119899 minus 120572)
119863120572
119891 (119905) = 119868119899minus120572
119886119863119899
lowast119891 (119905) 119899 = 1 2 (3)
We notice that the definition of time-fractional derivativeof a function 119891(119905) at 119905 = 119905
119899involves an integration and
calculating time-fractional derivative that requires all the pasthistory that is all the values of 119891(119905) from 119905 = 0 to 119905 =
119905119899 For the concept of fractional derivative we will adopt
Caputorsquos definition which is a modification of the Riemann-Liouville definition and has the advantage of dealing properlywith initial value problems The following remark addressessome of the main properties of the fractional derivatives andintegrals (see [8 29ndash31])
Remark 2 Let 120573 120574 isin R+ and 120572 isin (0 1) Then
(i) if 119868120573119886
1198711
rarr 1198711 and 119891(119905) isin 119871
1 then 119868120573
119886119868120574
119886119891(119905) =
119868120573+120574
119886119891(119905)
(ii) lim120573rarr119899
119868120573
119886119891(119909) = 119868
119899
119886119891(119905) uniformly on [119886 119887] 119899 = 1 2
3 where 1198681119886119891(119905) = int
119905
0119891(119904)119889119904
(iii) lim120573rarr0
119868120573
119886119891(119905) = 119891(119905) weakly
(iv) if 119891(119905) is absolutely continuous on [119886 119887] thenlim120572rarr1
119863120572
lowast119891(119905) = 119889119891(119905)119889119905
(v) thus 119863120572
lowast119891(119905) = (119889119889119905)119868
1minus120572
lowast119891(119905) (Riemann-Liouville
sense) and119863120572119891(119905) = 119868
1minus120572
lowast(119889119889119905)119891(119905) (Caputo sense)
The generalized mean value theorem and another propertyare defined in the following remark [32]
Remark 3
(i) Suppose that 119891(119905) isin 119862[119886 119887] and 119863120572
lowast119891(119905) isin 119862(119886 119887] for
0 lt 120572 le 1 and then we have
119891 (119905) = 119891 (119886) +
1
Γ (120572)
119863120572
lowast119891 (120585) (119905 minus 119886)
120572
with 119886 lt 120585 lt 119905 forall119905 isin (119886 119887]
(4)
(ii) If (i) holds and119863120572lowast119891(119905) ge 0 for all 119905 isin [119886 119887] then 119891(119905)
is nondecreasing for each 119905 isin [119886 119887] If 119863120572lowast119891(119905) le 0
for all 119905 isin [119886 119887] then 119891(119905) is nonincreasing for each119905 isin [119886 119887]
We next provide a class of fractional-order differentialmodels to describe the dynamics of tumour-immune systeminteractions
2 Fractional Model of Tumor-Immune System
Immune system is one of the most fascinating schemes fromthe point of view of biology and mathematics The immunesystem is complex intricate and interesting It is known to bemultifunctional andmultipathway somost immune effectorsdo more than one job Also each function of the immunesystem is typically done by more than one effector whichmakes it more robustThe reason of using FODEs is that theyare naturally related to systems with memory which exists intumor-immune interactions
Ordinary and delay differential equations have long beenused in modeling cancer phenomena [33ndash37] but fractional-order differential equations have short history in modelingsuch systems with memory The authors in [1] used a systemof fractional-order differential equations inmodeling cancer-immune system interactionThemodel includes two immuneeffectors 119864
1(119905) 1198642(119905) (such as cytotoxic T cells and natural
killer cells) interacting with the cancer cells 119879(119905) with aHolling function of type III (Holling type III describes asituation inwhich the number of prey consumed per predatorinitially rises slowly as the density of prey increases but thenlevels off with further increase in prey density In other wordsthe response of predators to prey is depressed at low preydensity then levels off with further increase in prey density)The model takes the form
119863120572
119879 = 119886119879 minus 11990311198791198641minus 11990321198791198642
119863120572
1198641= minus11988911198641+
11987921198641
1198792+ 1198961
0 lt 120572 le 1
119863120572
1198642= minus11988921198642+
11987921198642
1198792+ 1198961
(5)
where 119879 equiv 119879(119905) 1198641equiv 1198641(119905) 1198642equiv 1198642(119905) and 119886 119903
1 1199032 1198891
1198892 1198961 and 119896
2are positive constants The interaction terms
in the second and third equations of model (5) satisfy thecrossreactivity property of the immune system It has beenassumed that (119889
11198961(1 minus 119889
1)) ≪ (119889
21198962(1 minus 119889
2)) to avoid
Abstract and Applied Analysis 3
the nonbiological interior solution where both immuneeffectors coexist The equilibrium points of the system (5) are
E0= (0 0 0) E
1= (radic
11988911198961
(1 minus 1198891)
119886
1199031
0)
E2= (radic
11988921198962
(1 minus 1198892)
0
119886
1199032
)
(6)
The first equilibrium E0is the nave the second E
1is the
memory and the third E2is endemic according to the value
of the tumor size Stability analysis shows that the nave state isunstable However thememory state is locally asymptoticallystable if 119889
1lt 1198892and 119889
1lt 1 While the endemic state is
locally asymptotically stable if 1198892lt 1198891and 119889
2lt 1 there
is bifurcation at 1198891= 1 The stability of the memory state
depends on the value of one parameter namely the immuneeffector death rate
Now we modify model (5) to include three populationsof the activated immune-system cells 119864(119905) the tumor cells119879(119905) and the concentration of IL-2 in the single tumor-sitecompartment 119868
119871(119905) We consider the classic bilinear model
that includes Holling function of type I and external effectorcells 119904
1and input of IL-2 119904
2 (holling Type I is a linear
relationship where the predator is able to keep up withincreasing density of prey by eating them in direct proportionto their abundance in the environment If they eat 10 of theprey at low density they continue to eat 10 of them at highdensities) The interactions of the three populations are thengoverned by the fractional-order differential model
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864 + 1199013119864119868119871
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879 0 lt 120572
119894le 1 119894 = 1 2 3
1198631205723119868119871= 1199042+ 1199017119864119879 minus 119901
8119868119871
(7)
with initial conditions 119864(0) = 1198640 119879(0) = 119879
0 and 119868
119871(0) =
1198681198710
The parameter 1199011is the antigenicity rate of the tumor
(immune response to the appearance of the tumor) and 1199041
is the external source of the effector cells with rate of death1199012 The parameter 119901
4incorporates both multiplication and
death of tumor cells The maximal carrying capacity of thebiological environment for tumor cell is 119901minus1
5 1199013is considered
as the cooperation rate of effector cells with interleukin-2parameter 119901
6is the rate of tumor cells 119901
7is the competition
rate between the effector cells and the tumor cells Externalinput of IL-2 into the system is 119904
2 and the rate loss parameter
of IL-2 cells is 1199018
In the absence of immunotherapy with IL-2 we have
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864
0 lt 120572119894le 1 119894 = 1 2
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879
(8)
To ease the analysis and stability of the steady states withmeaningful parameters and minimize sensitivity (or robust-ness) of themodel we nondimensionalize the bilinear system(8) by taking the rescaling
119909 =
119864
1198640
119910 =
119879
1198790
120596 =
11990111198790
1199051199041198640
120579 =
1199012
119905119904
120590 =
1199041
1199051199041198640
119886 =
1199014
119905119904
119887 = 11990151198790 1 =
11990161198640
119905119904
120591 = 119905119904119905
(9)
Therefore after the previous substitution into (8) and replac-ing 120591 by 119905 the model becomes
1198631205721119909 = 120590 + 120596119909119910 minus 120579119909
1198631205722119910 = 119886119910 (1 minus 119887119910) minus 119909119910
(10)
21 Equilibria and Local Stability of Model (10) The steadystates of the reduced model (10) are again the intersection ofthe null clines 1198631205721119909 = 0 1198631205722119910 = 0 If 119910 = 0 the tumor-free equilibrium is atE
0= (119909 119910) = (120590120579 0) This steady state
is always exist since 120590120579 gt 0 From the analysis it is easyto prove that the tumor-free equilibriumE
0= (120590120579 0) of the
model (10) is asymptotically stable if threshold parameter (theminimum tumor-clearance parameter) R
0= 119886120579120590 lt 1 and
unstable ifR0gt 1
However if119910 = 0 the steady states are obtained by solving
1205961198861198871199102
minus 119886 (120596 + 120579119887) 119910 + 119886120579 minus 120590 = 0 (11)
In this case we have two endemic equilibria E1and E
2
E1= (1199091 1199101) where 119909
1=
minus119886 (119887120579 minus 120596) minus radicΔ
2120596
1199101=
119886 (119887120579 + 120596) + radicΔ
2119886119887120596
E2= (1199092 1199102) where 119909
2=
minus119886 (119887120579 minus 120596) + radicΔ
2120596
1199102=
119886 (119887120579 + 120596) minus radicΔ
2119886119887120596
(12)
with Δ = 1198862(119887120579 minus 120596)
2
+ 4120590120596119886119887 gt 0 The Jacobian matrix ofthe system (10) at the endemic equilibriumE
1is
119869 (E1) = (
1205961199101minus 120579 120596119909
1
minus1199101
119886 minus 21198861198871199101minus 1199091
) (13)
4 Abstract and Applied Analysis
Proposition 4 Assume that the endemic equilibrium E1
exists and has nonnegative coordinates If R0= 119886120579120590 lt 1
then tr(119869(E1)) gt 0 and E
1is unstable
Proof Since
tr (119869 (E1)) =
1205962minus 120596 (119886119887 + 119887120579) minus 119886119887
2120579
2119887120596
+
120596 minus 119886119887
2119886119887120596
radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(14)
then inequality tr(119869(E1)) gt 0 is true if
119886 [1205962
minus 120596 (119886119887 + 119887120579) minus 1198861198872
120579]
gt (119886119887 minus 120596)radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(15)
Therefore when 119886120579 lt 120590 we have1205962minus120596119887(119886+120579)minus1198861205791198872 gt 0 andhence both sides of the inequality are positiveTherefore if theequilibrium pointE
1exists and has nonnegative coordinates
then tr(119869(E1)) gt 0 and the point (E
1) is unstable whenever
R0= 119886120579120590 lt 1
Similarly we arrive at the following proposition
Proposition 5 If the point E2exists and has nonnegative
coordinates then it is asymptotically stable
Wenext examine a fractional-order differential model forHIV infection of CD4+ T cells
3 Fractional Model of HIV Infection ofCD4+ T Cells
As mentioned before many mathematical models havebeen developed to describe the immunological response toinfection with human immunodeficiency virus (HIV) usingordinary differential models (see eg [38 39]) and delaydifferential models [40 41] In this section we extend theanalysis and replace the original system of Perelson et al [39]by an equivalent fractional-order differential model of HIVinfection of CD4+ T cells that consists of three equations
1198631205721119867 = 119904 minus 120583
119867119867 + 119903119867(1 minus
119867 + 119868
119867max) minus 1198961119881119867
1198631205722119868 = 1198961015840
1119881119867 minus 120583
119868119868
1198631205723119881 = 119872120583
119887119868 minus 1198961119881119867 minus 120583
119881119881
05 lt 120572119894le 1 119894 = 1 2 3
(16)
Here119867 = 119867(119905) represents the concentration of healthy CD4+T cells at time 119905 119868 = 119868(119905) represents the concentration ofinfected CD4+ T cells and 119881 = 119881(119905) is the concentrationof free HIV at time 119905 In this system the logistic growth of
the healthy CD4+ T-cells is (1 minus ((119867 + 119868)119867max)) and theproliferation of infected cells is neglected The parameter 119904 isthe source of CD4+ T cells from precursors 120583
119867is the natural
death rate of CD4+ T cells (120583119867119867max gt 119904 cf [39 page 85])
119903 is their growth rate (thus 119903 gt 120583119867in general) and 119867max is
their carrying capacity The parameter 1198961represents the rate
of infection of T cells with free virus 11989610158401is the rate at which
infected cells become actively infected 120583119868is a blanket death
term for infected cells to reflect the assumption that we do notinitially know whether the cells die naturally or by burstingIn addition 120583
119887is the lytic death rate for infected cells Since
119872 viral particles are released by each lysing cell this term ismultiplied by the parameter119872 to represent the source for freevirus (assuming a one-time initial infection) Finally120583
119881is the
loss rate of virus The initial conditions for infection by freevirus are119867(0) = 119867
0 119868(0) = 119868
0 and 119881(0) = 119881
0
Theorem 6 According to Remark 3 and the fact that 119904 ge 0then the system (16) has a unique solution (119867 119868 119881)
119879 whichremains in R3
+and bounded by119867max [11]
31 Equilibria and Local Stability of Model (16) To evaluatethe equilibrium points of system (16) we put 1198631205721119867(119905) =
1198631205722119868(119905) = 119863
1205723119881(119905) = 0 Then the infection-free equilibrium
point (uninfected steady state) is Elowast0
= (1198670 0 0) and
endemic equilibrium point (infected steady state) is Elowast+
=
(119867lowast 119868lowast 119881lowast) where
119867lowast
=
120583119881120583119868
1198961015840
1119872120583119887minus 1198961120583119868
119868lowast
=
1198961015840
1119867lowast119881lowast
120583119868
119881lowast
=
120583119868[(119904 + (119903 minus 120583
119867)119867lowast)119867max minus 119903119867
lowast2
]
119867lowast[1198961015840
1119903119867lowastminus 1198961120583119868119867max]
(17)
The Jacobian matrix 119869(Elowast0) for system (16) evaluated at the
uninfected steady state Elowast0is then given by
119869 (Elowast
0)
= (
minus120583119867+ 119903 minus 2119903119867
0
119867max
minus1199031198670
119867maxminus11989611198670
0 minus120583119868
1198961015840
11198670
0 119872120583119887
minus (11989611198670+ 120583119881)
)
(18)
Let us introduce the following definition and assumption toease the analysis
Definition 7 The threshold parameter Rlowast0(the minimum
infection-free parameter) is the parameter that has theproperty that ifRlowast
0lt 1 then the endemic infected state does
not exist while ifRlowast0gt 1 the endemic infected state persists
where
Rlowast
0=
1198961015840
11205831198871198721198670
120583119868(120583119881+ 11989611198670)
(19)
Abstract and Applied Analysis 5
We also assume that
1198670=
119903 minus 120583119867+ [(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max]12
2119903119867minus1
max
(20)
The uninfected steady state is asymptotically stable if all of theeigenvalues120582 of the Jacobianmatrix 119869(Elowast
0) given by (18) have
negative real parts The characteristic equation det(119869(E0) minus
119868) = 0 becomes
(120582 + 120583119867minus 119903 +
21199031198670
119867max) (1205822
+ 119861120582 + 119862) = 0 (21)
where 119861 = 120583119868+1198961198681198670+120583119881and119862 = 120583
119868(11989611198670+120583119881)minus1198961015840
11205831198871198721198670
Hence the three roots of the characteristic equation (21) are
1205821= minus120583119867+ 119903 minus
1199031198670
119867maxequiv minusradic(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max lt 0
12058223
=
1
2
[minus119861 plusmn radic1198612minus 4119862]
(22)
Proposition 8 IfRlowast0equiv (1198961015840
11205831198871198721198670)120583119868(120583119881+11989611198670) lt 1 then
119862 gt 0 and the three roots of the characteristic equation (21)willhave negative real parts
Corollary 9 In case of uninfected steady state Elowast0 one has
three cases
(i) If Rlowast0
lt 0 the uninfected state is asymptoticallystable and the infected steady state E
+does not exist
(unphysical)
(ii) IfRlowast0= 1 then 119862 = 0 and from (21) implies that one
eigenvalue 120582 = 0 and the remaining two eigenvalueshave negative real parts The uninfected and infectedsteady states collide and there is a transcritical bifur-cation
(iii) IfRlowast0gt 1 then 119862 lt 0 and thus at least one eigenvalue
will be positive real root Thus the uninfected stateE0is unstable and the endemically infected state Elowast
+
emerges
To study the local stability of the positive infected steadystates Elowast
+for Rlowast
0gt 1 we consider the linearized system of
(16) at Elowast+ The Jacobian matrix at Elowast
+becomes
119869 (Elowast
+) =
(
(
minus119871lowast
minus119903119867lowast
119867maxminus1198961119867lowast
1198961015840
1119881lowast
minus120583119868
1198961015840
1119867lowast
1198961119881lowast
119872120583119887
minus (1198961119867lowast+ 120583119881)
)
)
(23)
Here
119871lowast
= minus[120583119867minus 119903 + 119896
1119881lowast
+
119903 (2119867lowast+ 119868lowast)
119867max] (24)
Then the characteristic equation of the linearized system is
119875 (120582) = 1205823
+ 11988611205822
+ 1198862120582 + 1198863= 0 (25)
1198861= 120583119868+ 120583119881+ 1198961119867lowast
+ 119871lowast
1198862= 119871lowast
(120583119868+ 120583119881+ 1198961119867lowast
) + 120583119868(120583119881+ 1198961119867lowast
)
minus 1198962
1119867lowast
119881lowast
minus 1198961015840
1119867lowast
(119872120583119887minus
119903119881lowast
119867max)
1198863= 1198961015840
1119867lowast
[1198961119872120583119887119881lowast
minus
119903120583119881119881lowast
119867maxminus 119871lowast
119872120583119887]
+ 119871lowast
120583119868(120583119881+ 1198961119867lowast
) minus 1205831198681198962
1119867lowast
119881lowast
(26)
The infected steady state Elowast+is asymptotically stable if all of
the eigenvalues have negative real parts This occurs if andonly if Routh-Hurwitz conditions are satisfied that is 119886
1gt 0
1198863gt 0 and 119886
11198862gt 1198863
4 Implicit Eulerrsquos Scheme for FODEs
Since most of the fractional-order differential equations donot have exact analytic solutions approximation and numer-ical techniques must be used Several numerical methodshave been proposed to solve the fractional-order differentialequations [18 42 43] In additionmost of resulting biologicalsystems are stiff (One definition of the stiffness is that theglobal accuracy of the numerical solution is determined bystability rather than local error and implicit methods aremore appropriate for it) The stiffness often appears due tothe differences in speed between the fastest and slowestcomponents of the solutions and stability constraints Inaddition the state variables of these types of models arevery sensitive to small perturbations (or changes) in theparameters occur in the model Therefore efficient use of areliable numerical method that based in general on implicitformulae for dealing with stiff problems is necessary
Consider biological models in the form of a system ofFODEs of the form
119863120572
119883(119905) = 119865 (119905 119883 (119905) P) 119905 isin [0 119879] 0 lt 120572 le 1
119883 (0) = 1198830
(27)
Here 119883(119905) = [1199091(119905) 1199092(119905) 119909
119899(119905)]119879 P is the set of param-
eters appear in themodel and119865(119905 119883(119905)) satisfies the Lipschitzcondition
119865 (119905 119883 (119905) P) minus 119865 (119905 119884 (119905) P) le 119870 119883 (119905) minus 119884 (119905)
119870 gt 0
(28)
where 119884(119905) is the solution of the perturbed system
Theorem10 Problem (27) has a unique solution provided thatthe Lipschitz condition (28) is satisfied and119870119879
120572Γ(120572 + 1) lt 1
Proof Using the definitions of Section 1 we can apply afractional integral operator to the differential equation (27)
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Abstract and Applied Analysis
We first give the definition of fractional-order integrationand fractional-order differentiation [29 30]There are severalapproaches to the generalization of the notion of differenti-ation to fractional orders for example Riemann-LiouvilleCaputo and Generalized Functions approaches Let 1198711 =
1198711[119886 119887] be the class of Lebesgue integrable functions on [119886 119887]
119886 lt 119887 lt infin
Definition 1 The fractional integral (or the Riemann-Liouville integral) of order 120573 isin R+ of the function 119891(119905) 119905 gt 0
(119891 R+ rarr R) is defined by
119868120573
119886119891 (119905) =
1
Γ (120573)
int
119905
119886
(119905 minus 119904)120573minus1
119891 (119904) 119889119904 119905 gt 0 (1)
The fractional derivative of order 120572 isin (119899 minus 1 119899) of 119891(119905) isdefined by two (nonequivalent) ways
(i) Riemann-Liouville fractional derivative take frac-tional integral of order (119899 minus 120572) and then take 119899thderivative as follows
119863120572
lowast119891 (119905) = 119863
119899
lowast119868119899minus120572
119886119891 (119905) 119863
119899
lowast=
119889119899
119889119905119899 119899 = 1 2
(2)
(ii) Caputo-fractional derivative take 119899th derivative andthen take a fractional integral of order (119899 minus 120572)
119863120572
119891 (119905) = 119868119899minus120572
119886119863119899
lowast119891 (119905) 119899 = 1 2 (3)
We notice that the definition of time-fractional derivativeof a function 119891(119905) at 119905 = 119905
119899involves an integration and
calculating time-fractional derivative that requires all the pasthistory that is all the values of 119891(119905) from 119905 = 0 to 119905 =
119905119899 For the concept of fractional derivative we will adopt
Caputorsquos definition which is a modification of the Riemann-Liouville definition and has the advantage of dealing properlywith initial value problems The following remark addressessome of the main properties of the fractional derivatives andintegrals (see [8 29ndash31])
Remark 2 Let 120573 120574 isin R+ and 120572 isin (0 1) Then
(i) if 119868120573119886
1198711
rarr 1198711 and 119891(119905) isin 119871
1 then 119868120573
119886119868120574
119886119891(119905) =
119868120573+120574
119886119891(119905)
(ii) lim120573rarr119899
119868120573
119886119891(119909) = 119868
119899
119886119891(119905) uniformly on [119886 119887] 119899 = 1 2
3 where 1198681119886119891(119905) = int
119905
0119891(119904)119889119904
(iii) lim120573rarr0
119868120573
119886119891(119905) = 119891(119905) weakly
(iv) if 119891(119905) is absolutely continuous on [119886 119887] thenlim120572rarr1
119863120572
lowast119891(119905) = 119889119891(119905)119889119905
(v) thus 119863120572
lowast119891(119905) = (119889119889119905)119868
1minus120572
lowast119891(119905) (Riemann-Liouville
sense) and119863120572119891(119905) = 119868
1minus120572
lowast(119889119889119905)119891(119905) (Caputo sense)
The generalized mean value theorem and another propertyare defined in the following remark [32]
Remark 3
(i) Suppose that 119891(119905) isin 119862[119886 119887] and 119863120572
lowast119891(119905) isin 119862(119886 119887] for
0 lt 120572 le 1 and then we have
119891 (119905) = 119891 (119886) +
1
Γ (120572)
119863120572
lowast119891 (120585) (119905 minus 119886)
120572
with 119886 lt 120585 lt 119905 forall119905 isin (119886 119887]
(4)
(ii) If (i) holds and119863120572lowast119891(119905) ge 0 for all 119905 isin [119886 119887] then 119891(119905)
is nondecreasing for each 119905 isin [119886 119887] If 119863120572lowast119891(119905) le 0
for all 119905 isin [119886 119887] then 119891(119905) is nonincreasing for each119905 isin [119886 119887]
We next provide a class of fractional-order differentialmodels to describe the dynamics of tumour-immune systeminteractions
2 Fractional Model of Tumor-Immune System
Immune system is one of the most fascinating schemes fromthe point of view of biology and mathematics The immunesystem is complex intricate and interesting It is known to bemultifunctional andmultipathway somost immune effectorsdo more than one job Also each function of the immunesystem is typically done by more than one effector whichmakes it more robustThe reason of using FODEs is that theyare naturally related to systems with memory which exists intumor-immune interactions
Ordinary and delay differential equations have long beenused in modeling cancer phenomena [33ndash37] but fractional-order differential equations have short history in modelingsuch systems with memory The authors in [1] used a systemof fractional-order differential equations inmodeling cancer-immune system interactionThemodel includes two immuneeffectors 119864
1(119905) 1198642(119905) (such as cytotoxic T cells and natural
killer cells) interacting with the cancer cells 119879(119905) with aHolling function of type III (Holling type III describes asituation inwhich the number of prey consumed per predatorinitially rises slowly as the density of prey increases but thenlevels off with further increase in prey density In other wordsthe response of predators to prey is depressed at low preydensity then levels off with further increase in prey density)The model takes the form
119863120572
119879 = 119886119879 minus 11990311198791198641minus 11990321198791198642
119863120572
1198641= minus11988911198641+
11987921198641
1198792+ 1198961
0 lt 120572 le 1
119863120572
1198642= minus11988921198642+
11987921198642
1198792+ 1198961
(5)
where 119879 equiv 119879(119905) 1198641equiv 1198641(119905) 1198642equiv 1198642(119905) and 119886 119903
1 1199032 1198891
1198892 1198961 and 119896
2are positive constants The interaction terms
in the second and third equations of model (5) satisfy thecrossreactivity property of the immune system It has beenassumed that (119889
11198961(1 minus 119889
1)) ≪ (119889
21198962(1 minus 119889
2)) to avoid
Abstract and Applied Analysis 3
the nonbiological interior solution where both immuneeffectors coexist The equilibrium points of the system (5) are
E0= (0 0 0) E
1= (radic
11988911198961
(1 minus 1198891)
119886
1199031
0)
E2= (radic
11988921198962
(1 minus 1198892)
0
119886
1199032
)
(6)
The first equilibrium E0is the nave the second E
1is the
memory and the third E2is endemic according to the value
of the tumor size Stability analysis shows that the nave state isunstable However thememory state is locally asymptoticallystable if 119889
1lt 1198892and 119889
1lt 1 While the endemic state is
locally asymptotically stable if 1198892lt 1198891and 119889
2lt 1 there
is bifurcation at 1198891= 1 The stability of the memory state
depends on the value of one parameter namely the immuneeffector death rate
Now we modify model (5) to include three populationsof the activated immune-system cells 119864(119905) the tumor cells119879(119905) and the concentration of IL-2 in the single tumor-sitecompartment 119868
119871(119905) We consider the classic bilinear model
that includes Holling function of type I and external effectorcells 119904
1and input of IL-2 119904
2 (holling Type I is a linear
relationship where the predator is able to keep up withincreasing density of prey by eating them in direct proportionto their abundance in the environment If they eat 10 of theprey at low density they continue to eat 10 of them at highdensities) The interactions of the three populations are thengoverned by the fractional-order differential model
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864 + 1199013119864119868119871
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879 0 lt 120572
119894le 1 119894 = 1 2 3
1198631205723119868119871= 1199042+ 1199017119864119879 minus 119901
8119868119871
(7)
with initial conditions 119864(0) = 1198640 119879(0) = 119879
0 and 119868
119871(0) =
1198681198710
The parameter 1199011is the antigenicity rate of the tumor
(immune response to the appearance of the tumor) and 1199041
is the external source of the effector cells with rate of death1199012 The parameter 119901
4incorporates both multiplication and
death of tumor cells The maximal carrying capacity of thebiological environment for tumor cell is 119901minus1
5 1199013is considered
as the cooperation rate of effector cells with interleukin-2parameter 119901
6is the rate of tumor cells 119901
7is the competition
rate between the effector cells and the tumor cells Externalinput of IL-2 into the system is 119904
2 and the rate loss parameter
of IL-2 cells is 1199018
In the absence of immunotherapy with IL-2 we have
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864
0 lt 120572119894le 1 119894 = 1 2
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879
(8)
To ease the analysis and stability of the steady states withmeaningful parameters and minimize sensitivity (or robust-ness) of themodel we nondimensionalize the bilinear system(8) by taking the rescaling
119909 =
119864
1198640
119910 =
119879
1198790
120596 =
11990111198790
1199051199041198640
120579 =
1199012
119905119904
120590 =
1199041
1199051199041198640
119886 =
1199014
119905119904
119887 = 11990151198790 1 =
11990161198640
119905119904
120591 = 119905119904119905
(9)
Therefore after the previous substitution into (8) and replac-ing 120591 by 119905 the model becomes
1198631205721119909 = 120590 + 120596119909119910 minus 120579119909
1198631205722119910 = 119886119910 (1 minus 119887119910) minus 119909119910
(10)
21 Equilibria and Local Stability of Model (10) The steadystates of the reduced model (10) are again the intersection ofthe null clines 1198631205721119909 = 0 1198631205722119910 = 0 If 119910 = 0 the tumor-free equilibrium is atE
0= (119909 119910) = (120590120579 0) This steady state
is always exist since 120590120579 gt 0 From the analysis it is easyto prove that the tumor-free equilibriumE
0= (120590120579 0) of the
model (10) is asymptotically stable if threshold parameter (theminimum tumor-clearance parameter) R
0= 119886120579120590 lt 1 and
unstable ifR0gt 1
However if119910 = 0 the steady states are obtained by solving
1205961198861198871199102
minus 119886 (120596 + 120579119887) 119910 + 119886120579 minus 120590 = 0 (11)
In this case we have two endemic equilibria E1and E
2
E1= (1199091 1199101) where 119909
1=
minus119886 (119887120579 minus 120596) minus radicΔ
2120596
1199101=
119886 (119887120579 + 120596) + radicΔ
2119886119887120596
E2= (1199092 1199102) where 119909
2=
minus119886 (119887120579 minus 120596) + radicΔ
2120596
1199102=
119886 (119887120579 + 120596) minus radicΔ
2119886119887120596
(12)
with Δ = 1198862(119887120579 minus 120596)
2
+ 4120590120596119886119887 gt 0 The Jacobian matrix ofthe system (10) at the endemic equilibriumE
1is
119869 (E1) = (
1205961199101minus 120579 120596119909
1
minus1199101
119886 minus 21198861198871199101minus 1199091
) (13)
4 Abstract and Applied Analysis
Proposition 4 Assume that the endemic equilibrium E1
exists and has nonnegative coordinates If R0= 119886120579120590 lt 1
then tr(119869(E1)) gt 0 and E
1is unstable
Proof Since
tr (119869 (E1)) =
1205962minus 120596 (119886119887 + 119887120579) minus 119886119887
2120579
2119887120596
+
120596 minus 119886119887
2119886119887120596
radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(14)
then inequality tr(119869(E1)) gt 0 is true if
119886 [1205962
minus 120596 (119886119887 + 119887120579) minus 1198861198872
120579]
gt (119886119887 minus 120596)radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(15)
Therefore when 119886120579 lt 120590 we have1205962minus120596119887(119886+120579)minus1198861205791198872 gt 0 andhence both sides of the inequality are positiveTherefore if theequilibrium pointE
1exists and has nonnegative coordinates
then tr(119869(E1)) gt 0 and the point (E
1) is unstable whenever
R0= 119886120579120590 lt 1
Similarly we arrive at the following proposition
Proposition 5 If the point E2exists and has nonnegative
coordinates then it is asymptotically stable
Wenext examine a fractional-order differential model forHIV infection of CD4+ T cells
3 Fractional Model of HIV Infection ofCD4+ T Cells
As mentioned before many mathematical models havebeen developed to describe the immunological response toinfection with human immunodeficiency virus (HIV) usingordinary differential models (see eg [38 39]) and delaydifferential models [40 41] In this section we extend theanalysis and replace the original system of Perelson et al [39]by an equivalent fractional-order differential model of HIVinfection of CD4+ T cells that consists of three equations
1198631205721119867 = 119904 minus 120583
119867119867 + 119903119867(1 minus
119867 + 119868
119867max) minus 1198961119881119867
1198631205722119868 = 1198961015840
1119881119867 minus 120583
119868119868
1198631205723119881 = 119872120583
119887119868 minus 1198961119881119867 minus 120583
119881119881
05 lt 120572119894le 1 119894 = 1 2 3
(16)
Here119867 = 119867(119905) represents the concentration of healthy CD4+T cells at time 119905 119868 = 119868(119905) represents the concentration ofinfected CD4+ T cells and 119881 = 119881(119905) is the concentrationof free HIV at time 119905 In this system the logistic growth of
the healthy CD4+ T-cells is (1 minus ((119867 + 119868)119867max)) and theproliferation of infected cells is neglected The parameter 119904 isthe source of CD4+ T cells from precursors 120583
119867is the natural
death rate of CD4+ T cells (120583119867119867max gt 119904 cf [39 page 85])
119903 is their growth rate (thus 119903 gt 120583119867in general) and 119867max is
their carrying capacity The parameter 1198961represents the rate
of infection of T cells with free virus 11989610158401is the rate at which
infected cells become actively infected 120583119868is a blanket death
term for infected cells to reflect the assumption that we do notinitially know whether the cells die naturally or by burstingIn addition 120583
119887is the lytic death rate for infected cells Since
119872 viral particles are released by each lysing cell this term ismultiplied by the parameter119872 to represent the source for freevirus (assuming a one-time initial infection) Finally120583
119881is the
loss rate of virus The initial conditions for infection by freevirus are119867(0) = 119867
0 119868(0) = 119868
0 and 119881(0) = 119881
0
Theorem 6 According to Remark 3 and the fact that 119904 ge 0then the system (16) has a unique solution (119867 119868 119881)
119879 whichremains in R3
+and bounded by119867max [11]
31 Equilibria and Local Stability of Model (16) To evaluatethe equilibrium points of system (16) we put 1198631205721119867(119905) =
1198631205722119868(119905) = 119863
1205723119881(119905) = 0 Then the infection-free equilibrium
point (uninfected steady state) is Elowast0
= (1198670 0 0) and
endemic equilibrium point (infected steady state) is Elowast+
=
(119867lowast 119868lowast 119881lowast) where
119867lowast
=
120583119881120583119868
1198961015840
1119872120583119887minus 1198961120583119868
119868lowast
=
1198961015840
1119867lowast119881lowast
120583119868
119881lowast
=
120583119868[(119904 + (119903 minus 120583
119867)119867lowast)119867max minus 119903119867
lowast2
]
119867lowast[1198961015840
1119903119867lowastminus 1198961120583119868119867max]
(17)
The Jacobian matrix 119869(Elowast0) for system (16) evaluated at the
uninfected steady state Elowast0is then given by
119869 (Elowast
0)
= (
minus120583119867+ 119903 minus 2119903119867
0
119867max
minus1199031198670
119867maxminus11989611198670
0 minus120583119868
1198961015840
11198670
0 119872120583119887
minus (11989611198670+ 120583119881)
)
(18)
Let us introduce the following definition and assumption toease the analysis
Definition 7 The threshold parameter Rlowast0(the minimum
infection-free parameter) is the parameter that has theproperty that ifRlowast
0lt 1 then the endemic infected state does
not exist while ifRlowast0gt 1 the endemic infected state persists
where
Rlowast
0=
1198961015840
11205831198871198721198670
120583119868(120583119881+ 11989611198670)
(19)
Abstract and Applied Analysis 5
We also assume that
1198670=
119903 minus 120583119867+ [(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max]12
2119903119867minus1
max
(20)
The uninfected steady state is asymptotically stable if all of theeigenvalues120582 of the Jacobianmatrix 119869(Elowast
0) given by (18) have
negative real parts The characteristic equation det(119869(E0) minus
119868) = 0 becomes
(120582 + 120583119867minus 119903 +
21199031198670
119867max) (1205822
+ 119861120582 + 119862) = 0 (21)
where 119861 = 120583119868+1198961198681198670+120583119881and119862 = 120583
119868(11989611198670+120583119881)minus1198961015840
11205831198871198721198670
Hence the three roots of the characteristic equation (21) are
1205821= minus120583119867+ 119903 minus
1199031198670
119867maxequiv minusradic(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max lt 0
12058223
=
1
2
[minus119861 plusmn radic1198612minus 4119862]
(22)
Proposition 8 IfRlowast0equiv (1198961015840
11205831198871198721198670)120583119868(120583119881+11989611198670) lt 1 then
119862 gt 0 and the three roots of the characteristic equation (21)willhave negative real parts
Corollary 9 In case of uninfected steady state Elowast0 one has
three cases
(i) If Rlowast0
lt 0 the uninfected state is asymptoticallystable and the infected steady state E
+does not exist
(unphysical)
(ii) IfRlowast0= 1 then 119862 = 0 and from (21) implies that one
eigenvalue 120582 = 0 and the remaining two eigenvalueshave negative real parts The uninfected and infectedsteady states collide and there is a transcritical bifur-cation
(iii) IfRlowast0gt 1 then 119862 lt 0 and thus at least one eigenvalue
will be positive real root Thus the uninfected stateE0is unstable and the endemically infected state Elowast
+
emerges
To study the local stability of the positive infected steadystates Elowast
+for Rlowast
0gt 1 we consider the linearized system of
(16) at Elowast+ The Jacobian matrix at Elowast
+becomes
119869 (Elowast
+) =
(
(
minus119871lowast
minus119903119867lowast
119867maxminus1198961119867lowast
1198961015840
1119881lowast
minus120583119868
1198961015840
1119867lowast
1198961119881lowast
119872120583119887
minus (1198961119867lowast+ 120583119881)
)
)
(23)
Here
119871lowast
= minus[120583119867minus 119903 + 119896
1119881lowast
+
119903 (2119867lowast+ 119868lowast)
119867max] (24)
Then the characteristic equation of the linearized system is
119875 (120582) = 1205823
+ 11988611205822
+ 1198862120582 + 1198863= 0 (25)
1198861= 120583119868+ 120583119881+ 1198961119867lowast
+ 119871lowast
1198862= 119871lowast
(120583119868+ 120583119881+ 1198961119867lowast
) + 120583119868(120583119881+ 1198961119867lowast
)
minus 1198962
1119867lowast
119881lowast
minus 1198961015840
1119867lowast
(119872120583119887minus
119903119881lowast
119867max)
1198863= 1198961015840
1119867lowast
[1198961119872120583119887119881lowast
minus
119903120583119881119881lowast
119867maxminus 119871lowast
119872120583119887]
+ 119871lowast
120583119868(120583119881+ 1198961119867lowast
) minus 1205831198681198962
1119867lowast
119881lowast
(26)
The infected steady state Elowast+is asymptotically stable if all of
the eigenvalues have negative real parts This occurs if andonly if Routh-Hurwitz conditions are satisfied that is 119886
1gt 0
1198863gt 0 and 119886
11198862gt 1198863
4 Implicit Eulerrsquos Scheme for FODEs
Since most of the fractional-order differential equations donot have exact analytic solutions approximation and numer-ical techniques must be used Several numerical methodshave been proposed to solve the fractional-order differentialequations [18 42 43] In additionmost of resulting biologicalsystems are stiff (One definition of the stiffness is that theglobal accuracy of the numerical solution is determined bystability rather than local error and implicit methods aremore appropriate for it) The stiffness often appears due tothe differences in speed between the fastest and slowestcomponents of the solutions and stability constraints Inaddition the state variables of these types of models arevery sensitive to small perturbations (or changes) in theparameters occur in the model Therefore efficient use of areliable numerical method that based in general on implicitformulae for dealing with stiff problems is necessary
Consider biological models in the form of a system ofFODEs of the form
119863120572
119883(119905) = 119865 (119905 119883 (119905) P) 119905 isin [0 119879] 0 lt 120572 le 1
119883 (0) = 1198830
(27)
Here 119883(119905) = [1199091(119905) 1199092(119905) 119909
119899(119905)]119879 P is the set of param-
eters appear in themodel and119865(119905 119883(119905)) satisfies the Lipschitzcondition
119865 (119905 119883 (119905) P) minus 119865 (119905 119884 (119905) P) le 119870 119883 (119905) minus 119884 (119905)
119870 gt 0
(28)
where 119884(119905) is the solution of the perturbed system
Theorem10 Problem (27) has a unique solution provided thatthe Lipschitz condition (28) is satisfied and119870119879
120572Γ(120572 + 1) lt 1
Proof Using the definitions of Section 1 we can apply afractional integral operator to the differential equation (27)
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 3
the nonbiological interior solution where both immuneeffectors coexist The equilibrium points of the system (5) are
E0= (0 0 0) E
1= (radic
11988911198961
(1 minus 1198891)
119886
1199031
0)
E2= (radic
11988921198962
(1 minus 1198892)
0
119886
1199032
)
(6)
The first equilibrium E0is the nave the second E
1is the
memory and the third E2is endemic according to the value
of the tumor size Stability analysis shows that the nave state isunstable However thememory state is locally asymptoticallystable if 119889
1lt 1198892and 119889
1lt 1 While the endemic state is
locally asymptotically stable if 1198892lt 1198891and 119889
2lt 1 there
is bifurcation at 1198891= 1 The stability of the memory state
depends on the value of one parameter namely the immuneeffector death rate
Now we modify model (5) to include three populationsof the activated immune-system cells 119864(119905) the tumor cells119879(119905) and the concentration of IL-2 in the single tumor-sitecompartment 119868
119871(119905) We consider the classic bilinear model
that includes Holling function of type I and external effectorcells 119904
1and input of IL-2 119904
2 (holling Type I is a linear
relationship where the predator is able to keep up withincreasing density of prey by eating them in direct proportionto their abundance in the environment If they eat 10 of theprey at low density they continue to eat 10 of them at highdensities) The interactions of the three populations are thengoverned by the fractional-order differential model
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864 + 1199013119864119868119871
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879 0 lt 120572
119894le 1 119894 = 1 2 3
1198631205723119868119871= 1199042+ 1199017119864119879 minus 119901
8119868119871
(7)
with initial conditions 119864(0) = 1198640 119879(0) = 119879
0 and 119868
119871(0) =
1198681198710
The parameter 1199011is the antigenicity rate of the tumor
(immune response to the appearance of the tumor) and 1199041
is the external source of the effector cells with rate of death1199012 The parameter 119901
4incorporates both multiplication and
death of tumor cells The maximal carrying capacity of thebiological environment for tumor cell is 119901minus1
5 1199013is considered
as the cooperation rate of effector cells with interleukin-2parameter 119901
6is the rate of tumor cells 119901
7is the competition
rate between the effector cells and the tumor cells Externalinput of IL-2 into the system is 119904
2 and the rate loss parameter
of IL-2 cells is 1199018
In the absence of immunotherapy with IL-2 we have
1198631205721119864 = 1199041+ 1199011119864119879 minus 119901
2119864
0 lt 120572119894le 1 119894 = 1 2
1198631205722119879 = 119901
4119879 (1 minus 119901
5119879) minus 119901
6119864119879
(8)
To ease the analysis and stability of the steady states withmeaningful parameters and minimize sensitivity (or robust-ness) of themodel we nondimensionalize the bilinear system(8) by taking the rescaling
119909 =
119864
1198640
119910 =
119879
1198790
120596 =
11990111198790
1199051199041198640
120579 =
1199012
119905119904
120590 =
1199041
1199051199041198640
119886 =
1199014
119905119904
119887 = 11990151198790 1 =
11990161198640
119905119904
120591 = 119905119904119905
(9)
Therefore after the previous substitution into (8) and replac-ing 120591 by 119905 the model becomes
1198631205721119909 = 120590 + 120596119909119910 minus 120579119909
1198631205722119910 = 119886119910 (1 minus 119887119910) minus 119909119910
(10)
21 Equilibria and Local Stability of Model (10) The steadystates of the reduced model (10) are again the intersection ofthe null clines 1198631205721119909 = 0 1198631205722119910 = 0 If 119910 = 0 the tumor-free equilibrium is atE
0= (119909 119910) = (120590120579 0) This steady state
is always exist since 120590120579 gt 0 From the analysis it is easyto prove that the tumor-free equilibriumE
0= (120590120579 0) of the
model (10) is asymptotically stable if threshold parameter (theminimum tumor-clearance parameter) R
0= 119886120579120590 lt 1 and
unstable ifR0gt 1
However if119910 = 0 the steady states are obtained by solving
1205961198861198871199102
minus 119886 (120596 + 120579119887) 119910 + 119886120579 minus 120590 = 0 (11)
In this case we have two endemic equilibria E1and E
2
E1= (1199091 1199101) where 119909
1=
minus119886 (119887120579 minus 120596) minus radicΔ
2120596
1199101=
119886 (119887120579 + 120596) + radicΔ
2119886119887120596
E2= (1199092 1199102) where 119909
2=
minus119886 (119887120579 minus 120596) + radicΔ
2120596
1199102=
119886 (119887120579 + 120596) minus radicΔ
2119886119887120596
(12)
with Δ = 1198862(119887120579 minus 120596)
2
+ 4120590120596119886119887 gt 0 The Jacobian matrix ofthe system (10) at the endemic equilibriumE
1is
119869 (E1) = (
1205961199101minus 120579 120596119909
1
minus1199101
119886 minus 21198861198871199101minus 1199091
) (13)
4 Abstract and Applied Analysis
Proposition 4 Assume that the endemic equilibrium E1
exists and has nonnegative coordinates If R0= 119886120579120590 lt 1
then tr(119869(E1)) gt 0 and E
1is unstable
Proof Since
tr (119869 (E1)) =
1205962minus 120596 (119886119887 + 119887120579) minus 119886119887
2120579
2119887120596
+
120596 minus 119886119887
2119886119887120596
radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(14)
then inequality tr(119869(E1)) gt 0 is true if
119886 [1205962
minus 120596 (119886119887 + 119887120579) minus 1198861198872
120579]
gt (119886119887 minus 120596)radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(15)
Therefore when 119886120579 lt 120590 we have1205962minus120596119887(119886+120579)minus1198861205791198872 gt 0 andhence both sides of the inequality are positiveTherefore if theequilibrium pointE
1exists and has nonnegative coordinates
then tr(119869(E1)) gt 0 and the point (E
1) is unstable whenever
R0= 119886120579120590 lt 1
Similarly we arrive at the following proposition
Proposition 5 If the point E2exists and has nonnegative
coordinates then it is asymptotically stable
Wenext examine a fractional-order differential model forHIV infection of CD4+ T cells
3 Fractional Model of HIV Infection ofCD4+ T Cells
As mentioned before many mathematical models havebeen developed to describe the immunological response toinfection with human immunodeficiency virus (HIV) usingordinary differential models (see eg [38 39]) and delaydifferential models [40 41] In this section we extend theanalysis and replace the original system of Perelson et al [39]by an equivalent fractional-order differential model of HIVinfection of CD4+ T cells that consists of three equations
1198631205721119867 = 119904 minus 120583
119867119867 + 119903119867(1 minus
119867 + 119868
119867max) minus 1198961119881119867
1198631205722119868 = 1198961015840
1119881119867 minus 120583
119868119868
1198631205723119881 = 119872120583
119887119868 minus 1198961119881119867 minus 120583
119881119881
05 lt 120572119894le 1 119894 = 1 2 3
(16)
Here119867 = 119867(119905) represents the concentration of healthy CD4+T cells at time 119905 119868 = 119868(119905) represents the concentration ofinfected CD4+ T cells and 119881 = 119881(119905) is the concentrationof free HIV at time 119905 In this system the logistic growth of
the healthy CD4+ T-cells is (1 minus ((119867 + 119868)119867max)) and theproliferation of infected cells is neglected The parameter 119904 isthe source of CD4+ T cells from precursors 120583
119867is the natural
death rate of CD4+ T cells (120583119867119867max gt 119904 cf [39 page 85])
119903 is their growth rate (thus 119903 gt 120583119867in general) and 119867max is
their carrying capacity The parameter 1198961represents the rate
of infection of T cells with free virus 11989610158401is the rate at which
infected cells become actively infected 120583119868is a blanket death
term for infected cells to reflect the assumption that we do notinitially know whether the cells die naturally or by burstingIn addition 120583
119887is the lytic death rate for infected cells Since
119872 viral particles are released by each lysing cell this term ismultiplied by the parameter119872 to represent the source for freevirus (assuming a one-time initial infection) Finally120583
119881is the
loss rate of virus The initial conditions for infection by freevirus are119867(0) = 119867
0 119868(0) = 119868
0 and 119881(0) = 119881
0
Theorem 6 According to Remark 3 and the fact that 119904 ge 0then the system (16) has a unique solution (119867 119868 119881)
119879 whichremains in R3
+and bounded by119867max [11]
31 Equilibria and Local Stability of Model (16) To evaluatethe equilibrium points of system (16) we put 1198631205721119867(119905) =
1198631205722119868(119905) = 119863
1205723119881(119905) = 0 Then the infection-free equilibrium
point (uninfected steady state) is Elowast0
= (1198670 0 0) and
endemic equilibrium point (infected steady state) is Elowast+
=
(119867lowast 119868lowast 119881lowast) where
119867lowast
=
120583119881120583119868
1198961015840
1119872120583119887minus 1198961120583119868
119868lowast
=
1198961015840
1119867lowast119881lowast
120583119868
119881lowast
=
120583119868[(119904 + (119903 minus 120583
119867)119867lowast)119867max minus 119903119867
lowast2
]
119867lowast[1198961015840
1119903119867lowastminus 1198961120583119868119867max]
(17)
The Jacobian matrix 119869(Elowast0) for system (16) evaluated at the
uninfected steady state Elowast0is then given by
119869 (Elowast
0)
= (
minus120583119867+ 119903 minus 2119903119867
0
119867max
minus1199031198670
119867maxminus11989611198670
0 minus120583119868
1198961015840
11198670
0 119872120583119887
minus (11989611198670+ 120583119881)
)
(18)
Let us introduce the following definition and assumption toease the analysis
Definition 7 The threshold parameter Rlowast0(the minimum
infection-free parameter) is the parameter that has theproperty that ifRlowast
0lt 1 then the endemic infected state does
not exist while ifRlowast0gt 1 the endemic infected state persists
where
Rlowast
0=
1198961015840
11205831198871198721198670
120583119868(120583119881+ 11989611198670)
(19)
Abstract and Applied Analysis 5
We also assume that
1198670=
119903 minus 120583119867+ [(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max]12
2119903119867minus1
max
(20)
The uninfected steady state is asymptotically stable if all of theeigenvalues120582 of the Jacobianmatrix 119869(Elowast
0) given by (18) have
negative real parts The characteristic equation det(119869(E0) minus
119868) = 0 becomes
(120582 + 120583119867minus 119903 +
21199031198670
119867max) (1205822
+ 119861120582 + 119862) = 0 (21)
where 119861 = 120583119868+1198961198681198670+120583119881and119862 = 120583
119868(11989611198670+120583119881)minus1198961015840
11205831198871198721198670
Hence the three roots of the characteristic equation (21) are
1205821= minus120583119867+ 119903 minus
1199031198670
119867maxequiv minusradic(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max lt 0
12058223
=
1
2
[minus119861 plusmn radic1198612minus 4119862]
(22)
Proposition 8 IfRlowast0equiv (1198961015840
11205831198871198721198670)120583119868(120583119881+11989611198670) lt 1 then
119862 gt 0 and the three roots of the characteristic equation (21)willhave negative real parts
Corollary 9 In case of uninfected steady state Elowast0 one has
three cases
(i) If Rlowast0
lt 0 the uninfected state is asymptoticallystable and the infected steady state E
+does not exist
(unphysical)
(ii) IfRlowast0= 1 then 119862 = 0 and from (21) implies that one
eigenvalue 120582 = 0 and the remaining two eigenvalueshave negative real parts The uninfected and infectedsteady states collide and there is a transcritical bifur-cation
(iii) IfRlowast0gt 1 then 119862 lt 0 and thus at least one eigenvalue
will be positive real root Thus the uninfected stateE0is unstable and the endemically infected state Elowast
+
emerges
To study the local stability of the positive infected steadystates Elowast
+for Rlowast
0gt 1 we consider the linearized system of
(16) at Elowast+ The Jacobian matrix at Elowast
+becomes
119869 (Elowast
+) =
(
(
minus119871lowast
minus119903119867lowast
119867maxminus1198961119867lowast
1198961015840
1119881lowast
minus120583119868
1198961015840
1119867lowast
1198961119881lowast
119872120583119887
minus (1198961119867lowast+ 120583119881)
)
)
(23)
Here
119871lowast
= minus[120583119867minus 119903 + 119896
1119881lowast
+
119903 (2119867lowast+ 119868lowast)
119867max] (24)
Then the characteristic equation of the linearized system is
119875 (120582) = 1205823
+ 11988611205822
+ 1198862120582 + 1198863= 0 (25)
1198861= 120583119868+ 120583119881+ 1198961119867lowast
+ 119871lowast
1198862= 119871lowast
(120583119868+ 120583119881+ 1198961119867lowast
) + 120583119868(120583119881+ 1198961119867lowast
)
minus 1198962
1119867lowast
119881lowast
minus 1198961015840
1119867lowast
(119872120583119887minus
119903119881lowast
119867max)
1198863= 1198961015840
1119867lowast
[1198961119872120583119887119881lowast
minus
119903120583119881119881lowast
119867maxminus 119871lowast
119872120583119887]
+ 119871lowast
120583119868(120583119881+ 1198961119867lowast
) minus 1205831198681198962
1119867lowast
119881lowast
(26)
The infected steady state Elowast+is asymptotically stable if all of
the eigenvalues have negative real parts This occurs if andonly if Routh-Hurwitz conditions are satisfied that is 119886
1gt 0
1198863gt 0 and 119886
11198862gt 1198863
4 Implicit Eulerrsquos Scheme for FODEs
Since most of the fractional-order differential equations donot have exact analytic solutions approximation and numer-ical techniques must be used Several numerical methodshave been proposed to solve the fractional-order differentialequations [18 42 43] In additionmost of resulting biologicalsystems are stiff (One definition of the stiffness is that theglobal accuracy of the numerical solution is determined bystability rather than local error and implicit methods aremore appropriate for it) The stiffness often appears due tothe differences in speed between the fastest and slowestcomponents of the solutions and stability constraints Inaddition the state variables of these types of models arevery sensitive to small perturbations (or changes) in theparameters occur in the model Therefore efficient use of areliable numerical method that based in general on implicitformulae for dealing with stiff problems is necessary
Consider biological models in the form of a system ofFODEs of the form
119863120572
119883(119905) = 119865 (119905 119883 (119905) P) 119905 isin [0 119879] 0 lt 120572 le 1
119883 (0) = 1198830
(27)
Here 119883(119905) = [1199091(119905) 1199092(119905) 119909
119899(119905)]119879 P is the set of param-
eters appear in themodel and119865(119905 119883(119905)) satisfies the Lipschitzcondition
119865 (119905 119883 (119905) P) minus 119865 (119905 119884 (119905) P) le 119870 119883 (119905) minus 119884 (119905)
119870 gt 0
(28)
where 119884(119905) is the solution of the perturbed system
Theorem10 Problem (27) has a unique solution provided thatthe Lipschitz condition (28) is satisfied and119870119879
120572Γ(120572 + 1) lt 1
Proof Using the definitions of Section 1 we can apply afractional integral operator to the differential equation (27)
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Abstract and Applied Analysis
Proposition 4 Assume that the endemic equilibrium E1
exists and has nonnegative coordinates If R0= 119886120579120590 lt 1
then tr(119869(E1)) gt 0 and E
1is unstable
Proof Since
tr (119869 (E1)) =
1205962minus 120596 (119886119887 + 119887120579) minus 119886119887
2120579
2119887120596
+
120596 minus 119886119887
2119886119887120596
radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(14)
then inequality tr(119869(E1)) gt 0 is true if
119886 [1205962
minus 120596 (119886119887 + 119887120579) minus 1198861198872
120579]
gt (119886119887 minus 120596)radic1198862(119887120579 + 120596)
2
minus 4119886119887120596 (119886120579 minus 120590)
(15)
Therefore when 119886120579 lt 120590 we have1205962minus120596119887(119886+120579)minus1198861205791198872 gt 0 andhence both sides of the inequality are positiveTherefore if theequilibrium pointE
1exists and has nonnegative coordinates
then tr(119869(E1)) gt 0 and the point (E
1) is unstable whenever
R0= 119886120579120590 lt 1
Similarly we arrive at the following proposition
Proposition 5 If the point E2exists and has nonnegative
coordinates then it is asymptotically stable
Wenext examine a fractional-order differential model forHIV infection of CD4+ T cells
3 Fractional Model of HIV Infection ofCD4+ T Cells
As mentioned before many mathematical models havebeen developed to describe the immunological response toinfection with human immunodeficiency virus (HIV) usingordinary differential models (see eg [38 39]) and delaydifferential models [40 41] In this section we extend theanalysis and replace the original system of Perelson et al [39]by an equivalent fractional-order differential model of HIVinfection of CD4+ T cells that consists of three equations
1198631205721119867 = 119904 minus 120583
119867119867 + 119903119867(1 minus
119867 + 119868
119867max) minus 1198961119881119867
1198631205722119868 = 1198961015840
1119881119867 minus 120583
119868119868
1198631205723119881 = 119872120583
119887119868 minus 1198961119881119867 minus 120583
119881119881
05 lt 120572119894le 1 119894 = 1 2 3
(16)
Here119867 = 119867(119905) represents the concentration of healthy CD4+T cells at time 119905 119868 = 119868(119905) represents the concentration ofinfected CD4+ T cells and 119881 = 119881(119905) is the concentrationof free HIV at time 119905 In this system the logistic growth of
the healthy CD4+ T-cells is (1 minus ((119867 + 119868)119867max)) and theproliferation of infected cells is neglected The parameter 119904 isthe source of CD4+ T cells from precursors 120583
119867is the natural
death rate of CD4+ T cells (120583119867119867max gt 119904 cf [39 page 85])
119903 is their growth rate (thus 119903 gt 120583119867in general) and 119867max is
their carrying capacity The parameter 1198961represents the rate
of infection of T cells with free virus 11989610158401is the rate at which
infected cells become actively infected 120583119868is a blanket death
term for infected cells to reflect the assumption that we do notinitially know whether the cells die naturally or by burstingIn addition 120583
119887is the lytic death rate for infected cells Since
119872 viral particles are released by each lysing cell this term ismultiplied by the parameter119872 to represent the source for freevirus (assuming a one-time initial infection) Finally120583
119881is the
loss rate of virus The initial conditions for infection by freevirus are119867(0) = 119867
0 119868(0) = 119868
0 and 119881(0) = 119881
0
Theorem 6 According to Remark 3 and the fact that 119904 ge 0then the system (16) has a unique solution (119867 119868 119881)
119879 whichremains in R3
+and bounded by119867max [11]
31 Equilibria and Local Stability of Model (16) To evaluatethe equilibrium points of system (16) we put 1198631205721119867(119905) =
1198631205722119868(119905) = 119863
1205723119881(119905) = 0 Then the infection-free equilibrium
point (uninfected steady state) is Elowast0
= (1198670 0 0) and
endemic equilibrium point (infected steady state) is Elowast+
=
(119867lowast 119868lowast 119881lowast) where
119867lowast
=
120583119881120583119868
1198961015840
1119872120583119887minus 1198961120583119868
119868lowast
=
1198961015840
1119867lowast119881lowast
120583119868
119881lowast
=
120583119868[(119904 + (119903 minus 120583
119867)119867lowast)119867max minus 119903119867
lowast2
]
119867lowast[1198961015840
1119903119867lowastminus 1198961120583119868119867max]
(17)
The Jacobian matrix 119869(Elowast0) for system (16) evaluated at the
uninfected steady state Elowast0is then given by
119869 (Elowast
0)
= (
minus120583119867+ 119903 minus 2119903119867
0
119867max
minus1199031198670
119867maxminus11989611198670
0 minus120583119868
1198961015840
11198670
0 119872120583119887
minus (11989611198670+ 120583119881)
)
(18)
Let us introduce the following definition and assumption toease the analysis
Definition 7 The threshold parameter Rlowast0(the minimum
infection-free parameter) is the parameter that has theproperty that ifRlowast
0lt 1 then the endemic infected state does
not exist while ifRlowast0gt 1 the endemic infected state persists
where
Rlowast
0=
1198961015840
11205831198871198721198670
120583119868(120583119881+ 11989611198670)
(19)
Abstract and Applied Analysis 5
We also assume that
1198670=
119903 minus 120583119867+ [(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max]12
2119903119867minus1
max
(20)
The uninfected steady state is asymptotically stable if all of theeigenvalues120582 of the Jacobianmatrix 119869(Elowast
0) given by (18) have
negative real parts The characteristic equation det(119869(E0) minus
119868) = 0 becomes
(120582 + 120583119867minus 119903 +
21199031198670
119867max) (1205822
+ 119861120582 + 119862) = 0 (21)
where 119861 = 120583119868+1198961198681198670+120583119881and119862 = 120583
119868(11989611198670+120583119881)minus1198961015840
11205831198871198721198670
Hence the three roots of the characteristic equation (21) are
1205821= minus120583119867+ 119903 minus
1199031198670
119867maxequiv minusradic(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max lt 0
12058223
=
1
2
[minus119861 plusmn radic1198612minus 4119862]
(22)
Proposition 8 IfRlowast0equiv (1198961015840
11205831198871198721198670)120583119868(120583119881+11989611198670) lt 1 then
119862 gt 0 and the three roots of the characteristic equation (21)willhave negative real parts
Corollary 9 In case of uninfected steady state Elowast0 one has
three cases
(i) If Rlowast0
lt 0 the uninfected state is asymptoticallystable and the infected steady state E
+does not exist
(unphysical)
(ii) IfRlowast0= 1 then 119862 = 0 and from (21) implies that one
eigenvalue 120582 = 0 and the remaining two eigenvalueshave negative real parts The uninfected and infectedsteady states collide and there is a transcritical bifur-cation
(iii) IfRlowast0gt 1 then 119862 lt 0 and thus at least one eigenvalue
will be positive real root Thus the uninfected stateE0is unstable and the endemically infected state Elowast
+
emerges
To study the local stability of the positive infected steadystates Elowast
+for Rlowast
0gt 1 we consider the linearized system of
(16) at Elowast+ The Jacobian matrix at Elowast
+becomes
119869 (Elowast
+) =
(
(
minus119871lowast
minus119903119867lowast
119867maxminus1198961119867lowast
1198961015840
1119881lowast
minus120583119868
1198961015840
1119867lowast
1198961119881lowast
119872120583119887
minus (1198961119867lowast+ 120583119881)
)
)
(23)
Here
119871lowast
= minus[120583119867minus 119903 + 119896
1119881lowast
+
119903 (2119867lowast+ 119868lowast)
119867max] (24)
Then the characteristic equation of the linearized system is
119875 (120582) = 1205823
+ 11988611205822
+ 1198862120582 + 1198863= 0 (25)
1198861= 120583119868+ 120583119881+ 1198961119867lowast
+ 119871lowast
1198862= 119871lowast
(120583119868+ 120583119881+ 1198961119867lowast
) + 120583119868(120583119881+ 1198961119867lowast
)
minus 1198962
1119867lowast
119881lowast
minus 1198961015840
1119867lowast
(119872120583119887minus
119903119881lowast
119867max)
1198863= 1198961015840
1119867lowast
[1198961119872120583119887119881lowast
minus
119903120583119881119881lowast
119867maxminus 119871lowast
119872120583119887]
+ 119871lowast
120583119868(120583119881+ 1198961119867lowast
) minus 1205831198681198962
1119867lowast
119881lowast
(26)
The infected steady state Elowast+is asymptotically stable if all of
the eigenvalues have negative real parts This occurs if andonly if Routh-Hurwitz conditions are satisfied that is 119886
1gt 0
1198863gt 0 and 119886
11198862gt 1198863
4 Implicit Eulerrsquos Scheme for FODEs
Since most of the fractional-order differential equations donot have exact analytic solutions approximation and numer-ical techniques must be used Several numerical methodshave been proposed to solve the fractional-order differentialequations [18 42 43] In additionmost of resulting biologicalsystems are stiff (One definition of the stiffness is that theglobal accuracy of the numerical solution is determined bystability rather than local error and implicit methods aremore appropriate for it) The stiffness often appears due tothe differences in speed between the fastest and slowestcomponents of the solutions and stability constraints Inaddition the state variables of these types of models arevery sensitive to small perturbations (or changes) in theparameters occur in the model Therefore efficient use of areliable numerical method that based in general on implicitformulae for dealing with stiff problems is necessary
Consider biological models in the form of a system ofFODEs of the form
119863120572
119883(119905) = 119865 (119905 119883 (119905) P) 119905 isin [0 119879] 0 lt 120572 le 1
119883 (0) = 1198830
(27)
Here 119883(119905) = [1199091(119905) 1199092(119905) 119909
119899(119905)]119879 P is the set of param-
eters appear in themodel and119865(119905 119883(119905)) satisfies the Lipschitzcondition
119865 (119905 119883 (119905) P) minus 119865 (119905 119884 (119905) P) le 119870 119883 (119905) minus 119884 (119905)
119870 gt 0
(28)
where 119884(119905) is the solution of the perturbed system
Theorem10 Problem (27) has a unique solution provided thatthe Lipschitz condition (28) is satisfied and119870119879
120572Γ(120572 + 1) lt 1
Proof Using the definitions of Section 1 we can apply afractional integral operator to the differential equation (27)
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 5
We also assume that
1198670=
119903 minus 120583119867+ [(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max]12
2119903119867minus1
max
(20)
The uninfected steady state is asymptotically stable if all of theeigenvalues120582 of the Jacobianmatrix 119869(Elowast
0) given by (18) have
negative real parts The characteristic equation det(119869(E0) minus
119868) = 0 becomes
(120582 + 120583119867minus 119903 +
21199031198670
119867max) (1205822
+ 119861120582 + 119862) = 0 (21)
where 119861 = 120583119868+1198961198681198670+120583119881and119862 = 120583
119868(11989611198670+120583119881)minus1198961015840
11205831198871198721198670
Hence the three roots of the characteristic equation (21) are
1205821= minus120583119867+ 119903 minus
1199031198670
119867maxequiv minusradic(119903 minus 120583
119867)2
+ 4119903119904119867minus1
max lt 0
12058223
=
1
2
[minus119861 plusmn radic1198612minus 4119862]
(22)
Proposition 8 IfRlowast0equiv (1198961015840
11205831198871198721198670)120583119868(120583119881+11989611198670) lt 1 then
119862 gt 0 and the three roots of the characteristic equation (21)willhave negative real parts
Corollary 9 In case of uninfected steady state Elowast0 one has
three cases
(i) If Rlowast0
lt 0 the uninfected state is asymptoticallystable and the infected steady state E
+does not exist
(unphysical)
(ii) IfRlowast0= 1 then 119862 = 0 and from (21) implies that one
eigenvalue 120582 = 0 and the remaining two eigenvalueshave negative real parts The uninfected and infectedsteady states collide and there is a transcritical bifur-cation
(iii) IfRlowast0gt 1 then 119862 lt 0 and thus at least one eigenvalue
will be positive real root Thus the uninfected stateE0is unstable and the endemically infected state Elowast
+
emerges
To study the local stability of the positive infected steadystates Elowast
+for Rlowast
0gt 1 we consider the linearized system of
(16) at Elowast+ The Jacobian matrix at Elowast
+becomes
119869 (Elowast
+) =
(
(
minus119871lowast
minus119903119867lowast
119867maxminus1198961119867lowast
1198961015840
1119881lowast
minus120583119868
1198961015840
1119867lowast
1198961119881lowast
119872120583119887
minus (1198961119867lowast+ 120583119881)
)
)
(23)
Here
119871lowast
= minus[120583119867minus 119903 + 119896
1119881lowast
+
119903 (2119867lowast+ 119868lowast)
119867max] (24)
Then the characteristic equation of the linearized system is
119875 (120582) = 1205823
+ 11988611205822
+ 1198862120582 + 1198863= 0 (25)
1198861= 120583119868+ 120583119881+ 1198961119867lowast
+ 119871lowast
1198862= 119871lowast
(120583119868+ 120583119881+ 1198961119867lowast
) + 120583119868(120583119881+ 1198961119867lowast
)
minus 1198962
1119867lowast
119881lowast
minus 1198961015840
1119867lowast
(119872120583119887minus
119903119881lowast
119867max)
1198863= 1198961015840
1119867lowast
[1198961119872120583119887119881lowast
minus
119903120583119881119881lowast
119867maxminus 119871lowast
119872120583119887]
+ 119871lowast
120583119868(120583119881+ 1198961119867lowast
) minus 1205831198681198962
1119867lowast
119881lowast
(26)
The infected steady state Elowast+is asymptotically stable if all of
the eigenvalues have negative real parts This occurs if andonly if Routh-Hurwitz conditions are satisfied that is 119886
1gt 0
1198863gt 0 and 119886
11198862gt 1198863
4 Implicit Eulerrsquos Scheme for FODEs
Since most of the fractional-order differential equations donot have exact analytic solutions approximation and numer-ical techniques must be used Several numerical methodshave been proposed to solve the fractional-order differentialequations [18 42 43] In additionmost of resulting biologicalsystems are stiff (One definition of the stiffness is that theglobal accuracy of the numerical solution is determined bystability rather than local error and implicit methods aremore appropriate for it) The stiffness often appears due tothe differences in speed between the fastest and slowestcomponents of the solutions and stability constraints Inaddition the state variables of these types of models arevery sensitive to small perturbations (or changes) in theparameters occur in the model Therefore efficient use of areliable numerical method that based in general on implicitformulae for dealing with stiff problems is necessary
Consider biological models in the form of a system ofFODEs of the form
119863120572
119883(119905) = 119865 (119905 119883 (119905) P) 119905 isin [0 119879] 0 lt 120572 le 1
119883 (0) = 1198830
(27)
Here 119883(119905) = [1199091(119905) 1199092(119905) 119909
119899(119905)]119879 P is the set of param-
eters appear in themodel and119865(119905 119883(119905)) satisfies the Lipschitzcondition
119865 (119905 119883 (119905) P) minus 119865 (119905 119884 (119905) P) le 119870 119883 (119905) minus 119884 (119905)
119870 gt 0
(28)
where 119884(119905) is the solution of the perturbed system
Theorem10 Problem (27) has a unique solution provided thatthe Lipschitz condition (28) is satisfied and119870119879
120572Γ(120572 + 1) lt 1
Proof Using the definitions of Section 1 we can apply afractional integral operator to the differential equation (27)
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Abstract and Applied Analysis
and incorporate the initial conditions thus converting theequation into the equivalent equation
119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904 (29)
which also is a Volterra equation of the second kind Definethe operatorL such that
L119883 (119905) = 119883 (0) +
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) 119889119904
(30)
Then we have
L119883 (119905) minusL119884 (119905)
le
1
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1
119865 (119904 119883 (119904) P) minus 119865 (119904 119884 (119904) P) 119889119904
le
119870
Γ (120572)
int
119905
0
(119905 minus 119904)120572minus1 sup119904isin[0119879]
|119883 (119904) minus 119884 (119904)| 119889119904
le
119870
Γ (120572)
119883 minus 119884int
119905
0
119904120572minus1
119889119904
le
119870
Γ (120572)
119883 minus 119884119879120572
(31)
Thus for119870119879120572Γ(120572 + 1) lt 1 we have
L119883 (119905) minusL119884 (119905) le 119883 minus 119884 (32)
This implies that our problem has a unique solution
However converted Volterra integral equation (29) iswith a weakly singular kernel such that a regularization isnot necessary any more It seems that there exists only a verysmall number of software packages for nonlinear Volterraequations In our case the kernel may not be continuousand therefore the classical numerical algorithms for theintegral part of (29) are unable to handle the solution of(27) Therefore we implement the implicit Eulerrsquos scheme toapproximate the fractional-order derivative
Given model (27) and mesh points T = 1199050 1199051 119905
119873
such that 1199050= 0 and 119905
119873= 119879 Then a discrete approximation
to the fractional derivative can be obtained by a simplequadrature formula using the Caputo fractional derivative(3) of order 120572 0 lt 120572 le 1 and using implicit Eulerrsquosapproximation as follows (see [44])
119863120572
lowast119909119894(119905119899)
=
1
Γ (1 minus 120572)
int
119905
0
119889119909119894(119904)
119889119904
(119905119899minus 119904)minus120572
119889119904
asymp
1
Γ (1 minus 120572)
119899
sum
119895=1
int
119895ℎ
(119895minus1)ℎ
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)] (119899ℎ minus 119904)minus120572
119889119904
=
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[
119909119895
119894minus 119909119895minus1
119894
ℎ
+ 119874 (ℎ)]
times [(119899 minus 119895 + 1)1minus120572
minus (119899 minus 119895)1minus120572
] ℎ1minus120572
=
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]
+
1
(1 minus 120572) Γ (1 minus 120572)
times
119899
sum
119895=1
[119909119895
119894minus 119909119895minus1
119894] [(119899 minus 119895 + 1)
1minus120572
minus (119899 minus 119895)1minus120572
]119874 (ℎ2minus120572
)
(33)
Setting
G (120572 ℎ) =
1
(1 minus 120572) Γ (1 minus 120572)
1
ℎ120572 120596
120572
119895= 1198951minus120572
minus (119895 minus 1)1minus120572
(where 120596120572
1= 1)
(34)
then the first-order approximation method for the compu-tation of Caputorsquos fractional derivative is then given by theexpression
119863120572
lowast119909119894(119905119899) = G (120572 ℎ)
119899
sum
119895=1
120596120572
119895(119909119899minus119895+1
119894minus 119909119899minus119895
119894) + 119874 (ℎ) (35)
From the analysis and numerical approximation we alsoarrive at the following proposition
Proposition 11 The presence of a fractional differential orderin a differential equation can lead to a notable increase inthe complexity of the observed behaviour and the solution iscontinuously depends on all the previous states
41 Stability and Convergence We here prove that thefractional-order implicit difference approximation (35) isunconditionally stable It follows then that the numericalsolution converges to the exact solution as ℎ rarr 0
In order to study the stability of the numerical methodlet us consider a test problem of linear scaler fractionaldifferential equation
119863120572
lowast119906 (119905) = 120588
0119906 (119905) + 120588
1 119880 (0) = 119880
0 (36)
such that 0 lt 120572 le 1 and 1205880lt 0 120588
1gt 0 are constants
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 7
0 20 40 60 80 1000
05
1
15
Time0 20 40 60 80 100
0
05
1
15
2
25
Time
0 20 40 60 80 1000
02
04
06
08
1
12
14
16
18
Time0 05 1 15 2 25
0
02
04
06
08
1
12
14
16
Tumor
T(t)
T(t)
T(t)
E1(t)
E1(t)
E1(t)
E2(t)
T(t)E1(t)E2(t)
T(t)E1(t)E2(t)
E2(t)
E2(t)
T(t)
E1(t)
E2(t)
120572 = 095
120572 = 075 120572 = 075
120572 = 075
Effec
tor c
ellsE1(t)E2(t)
Figure 1 Numerical simulations of the FODEs model (5) when 120572 = 075 and 120572 = 095 with (119886 = 1199031= 1199032= 1 119889
1= 03 119889
2= 07 119896
1= 03
1198962= 07) and when 119889
1= 07 119889
2= 03 The system converges to the stable steady statesE
1E2 The fractional derivative damps the oscillation
behavior
Theorem 12 The fully implicit numerical approximation (35)to test problem (36) for all 119905 ge 0 is consistent and uncon-ditionally stable
Proof We assume that the approximate solution of (36) is ofthe form 119906(119905
119899) asymp 119880119899equiv 120577119899 and then (36) can be reduced to
(1 minus
1205880
G120572ℎ
)120577119899
= 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) +
1205881
G120572ℎ
119899 ge 2
(37)
or
120577119899=
120577119899minus1
+ sum119899
119895=2120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) + 1205881G120572ℎ
(1 minus (1205880G120572ℎ))
119899 ge 2
(38)
Since (1 minus (1205880G120572ℎ)) ge 1 for allG
120572ℎ then
1205771le 1205770 (39)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) 119899 ge 2 (40)
Thus for 119899 = 2 the previous inequality implies
1205772le 1205771+ 120596(120572)
2(1205770minus 1205771) (41)
Using the relation (39) and the positivity of the coefficients1205962 we get
1205772le 1205771 (42)
Repeating the process we have from (40)
120577119899le 120577119899minus1
+
119899
sum
119895=2
120596(120572)
119895(120577119899minus119895
minus 120577119899minus119895+1
) le 120577119899minus1
(43)
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Abstract and Applied Analysis
0 50 100 1500
05
1
15
2
25
3
35
4
Time0 50 100 150
2
4
6
8
10
12
Time
0 05 1 15 2 25 3 35 40
2
4
6
8
10
12
Effec
tor c
ells
E(t)
Effector cells E(t)
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09
120572 = 07
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 2 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 01747 119903 = 0636 and 119887 = 0002 1205721= 1205722=
1 09 07The endemic stateE2is locally asymptotically stable whenR
0= 119886120579120590 gt 1The fractional derivative damps the oscillation behavior
since each term in the summation is negative Thus 120577119899
le
120577119899minus1
le 120577119899minus2
le sdot sdot sdot le 1205770 With the assumption that 120577
119899= |119880119899| le
1205770= |1198800| which entails 119880119899 le 119880
0 we have stability
Of course this numerical technique can be used both forlinear and for nonlinear problems and it may be extendedto multiterm FODEs For more details about stability andconvergence of the fractional Euler method we refer to [1945]
42 Numerical Simulations We employed the implicit Eulerrsquosscheme (35) to solve the resulting biological systems ofFODEs (5) (10) and (16) Interesting numerical simulationsof the fractional tumor-immune models (5) and (10) withstep size ℎ = 005 and 05 lt 120572 le 1 and parameters valuesgiven in the captions are displayed in Figures 1 2 and 3 The
equilibrium points (for infection-free and endemic cases) arethe same in both integer-order (when 120572 = 1) and fractional-order (120572 lt 1) models We notice that in the endemic steadystates the fractional-order derivative damps the oscillationbehavior From the graphs we can see that FODEs have richdynamics and are better descriptors of biological systemsthan traditional integer-order models
The numerical simulations for the HIV FODEmodel (16)(with parameter values given in the captions) are displayed inFigure 4We note that the solution of themodel with variousvalues of 120572 continuously depends on the time-fractionalderivative but arrives to the equilibriumpointsThe displayedsolutions in the figure confirm that the fractional order ofthe derivative plays the role of time delay in the system Weshould also note that although the equilibrium points arethe same for both integer-order and fractional-order models
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 9
025 03 035 04 045 05 055 06 065 07 0750
005
01
015
02
025
03
035
Effector cells E(t)
0 20 40 60 80 100 120 140 160 180 200025
03
035
04
045
05
055
06
065
07
075
Time
Effec
tor c
ells
E(t)
120572 = 1
120572 = 09
120572 = 08
120572 = 1
120572 = 09
120572 = 08
0 20 40 60 80 100 120 140 160 180 2000
005
01
015
02
025
03
035
Time
120572 = 1
120572 = 09
120572 = 08
Tum
or ce
llsT
(t)
Tum
or ce
llsT
(t)
Figure 3 Numerical simulations of the FODEs model (10) when 119904 = 01181 120596 = 01184 120579 = 03747 119903 = 1636 and 119887 = 0002 1205721= 1205722=
1 09 07 The infection-free steady state E0is locally asymptotically stable whenR
0= 119886120579120590 lt 1
the solution of the fractional order model tends to the fixedpoint over a longer period of time
5 Conclusions
In fact fractional-order differential equations are generaliza-tions of integer-order differential equations Using fractional-order differential equations can help us to reduce the errorsarising from the neglected parameters in modeling biologicalsystems withmemory and systems distributed parameters Inthis paper we presented a class of fractional-order differentialmodels of biological systems with memory to model theinteraction of immune system with tumor cells and withHIV infection of CD4+ T cells The models possess non-negative solutions as desired in any population dynamicsWe obtained the threshold parameterR
0that represents the
minimum tumor-clearance parameter orminimum infectionfree for each model
We provided unconditionally stable numerical techniqueusing the Caputo fractional derivative of order 120572 and implicitEulerrsquos approximation for the resulting system The numer-ical technique is suitable for stiff problems The solutionof the system at any time 119905
lowast is continuously depends onall the previous states at 119905 le 119905
lowast We have seen that thepresence of the fractional differential order leads to a notableincrease in the complexity of the observed behaviour and playthe role of time lag (or delay term) in ordinary differentialmodel We have obtained stability conditions for disease-free equilibrium and nonstability conditions for positiveequilibria We should also mention that one of the basicreasons of using fractional-order differential equations is thatfractional-order differential equations are at least as stableas their integer-order counterpart In addition the presenceof a fractional differential order in a differential equation canlead to a notable increase in the complexity of the observedbehaviour and the solution continuously depends on all
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Abstract and Applied Analysis
0 20 40 60 80 100 120 140 160 180 2000
200
400
600
800
1000
1200
Time0 20 40 60 80 100 120 140 160 180 200
0100200300400500600700
Time
0 20 40 60 80 100 120 140 160 180 2000
2
4
6
8
10
Time
Initi
al d
ensit
y of
HIV
RN
A
0 200 400 600 800 1000 1200
02004006008000
5
10
H(t)I(t)
V(t)120572 = 1
120572 = 09120572 = 07
120572 = 1
120572 = 09
120572 = 07Uni
nfec
ted
CD4+
T-ce
ll po
pulat
ion
size
120572 = 1
120572 = 09
120572 = 07
120572 = 1
120572 = 09120572 = 07
Infe
cted
CD
4+T-
cell
dens
ity
times104
times104
Figure 4 Numerical simulations of the FODEs model (16) when 119904 = 20 dayminus1mmminus3 120583119867= 002 dayminus1 120583
119868= 026 dayminus1 120583
119887= 024 dayminus1
120583119881= 24 dayminus1 119896
1= 24 times 10
minus5mm3 dayminus1 11989610158401= 2 times 10
minus5mm3 dayminus1 119903 = 03 dayminus1119867max = 1500mmminus3 and119872 = 1400 The trajectories ofthe system approach to the steady state Elowast
+
the previous states The analysis can be extended to othermodels related to the immune response systems
Acknowledgments
The work is funded by the UAE University NRF Project noNRF-7-20886 The author is grateful to Professor ChangpinLirsquos valuable suggestions and refereesrsquo constructive com-ments
References
[1] E Ahmed A Hashish and F A Rihan ldquoOn fractional ordercancer modelrdquo Journal of Fractional Calculus and AppliedAnalysis vol 3 no 2 pp 1ndash6 2012
[2] A A M Arafa S Z Rida and M Khalil ldquoFractional modelingdynamics of HIV and CD4+ T-cells during primary infectionrdquoNonlinear Biomedical Physics vol 6 no 1 article 1 2012
[3] K S Cole ldquoElectric conductance of biological systemsrdquo ColdSpring Harbor Symposia on Quantitative Biology pp 107ndash1161993
[4] AMA El-Sayed A EM El-Mesiry andHA A El-Saka ldquoOnthe fractional-order logistic equationrdquo Applied MathematicsLetters vol 20 no 7 pp 817ndash823 2007
[5] H Xu ldquoAnalytical approximations for a population growthmodel with fractional orderrdquo Communications in NonlinearScience and Numerical Simulation vol 14 no 5 pp 1978ndash19832009
[6] L Debnath ldquoRecent applications of fractional calculus toscience and engineeringrdquo International Journal of Mathematicsand Mathematical Sciences no 54 pp 3413ndash3442 2003
[7] A M A El-Sayed ldquoNonlinear functional-differential equationsof arbitrary ordersrdquo Nonlinear Analysis Theory Methods ampApplications vol 33 no 2 pp 181ndash186 1998
[8] R Hilfer Ed Applications of Fractional Calculus in PhysicsWorld Scientific River Edge NJ USA 2000
[9] G M Zaslavsky ldquoChaos fractional kinetics and anomaloustransportrdquo Physics Reports vol 371 no 6 pp 461ndash580 2002
[10] S B Yuste L Acedo and K Lindenberg ldquoReaction front in anA + B rarr C reaction-subdiffusion processrdquo Physical Review Evol 69 no 3 Article ID 036126 pp 1ndash36126 2004
[11] W Lin ldquoGlobal existence theory and chaos control of fractionaldifferential equationsrdquo Journal of Mathematical Analysis andApplications vol 332 no 1 pp 709ndash726 2007
[12] H Sheng Y Q Chen and T S Qiu Fractional Processes andFractional-Order Signal Processing Springer New York NYUSA 2012
[13] K Assaleh and W M Ahmad ldquoModeling of speech signalsusing fractional calculusrdquo in Proceedings of the 9th InternationalSymposium on Signal Processing and its Applications (ISSPA rsquo07)Sharjah United Arab Emirates February 2007
[14] Y Ferdi ldquoSome applications of fractional order calculus todesign digital filters for biomedical signal processingrdquo Journalof Mechanics in Medicine and Biology vol 12 no 2 Article ID12400088 13 pages 2012
[15] W-C Chen ldquoNonlinear dynamics and chaos in a fractional-order financial systemrdquo Chaos Solitons and Fractals vol 36 no5 pp 1305ndash1314 2008
[16] A Rocco and B J West ldquoFractional calculus and the evolutionof fractal phenomenardquo Physica A vol 265 no 3 pp 535ndash5461999
[17] V D Djordjevic J Jaric B Fabry J J Fredberg and DStamenovic ldquoFractional derivatives embody essential featuresof cell rheological behaviorrdquo Annals of Biomedical Engineeringvol 31 no 6 pp 692ndash699 2003
[18] K Diethelm N J Ford and A D Freed ldquoA predictor-correctorapproach for the numerical solution of fractional differentialequationsrdquoNonlinear Dynamics vol 29 no 1ndash4 pp 3ndash22 2002
[19] C Li and F Zeng ldquoThe finite difference methods for fractionalordinary differential equationsrdquo Numerical Functional Analysisand Optimization vol 34 no 2 pp 149ndash179 2013
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 11
[20] J C Trigeassou NMaamri J Sabatier andAOustaloup ldquoStatevariables and transients of fractional order differential systemsrdquoComputers ampMathematics with Applications vol 64 no 10 pp3117ndash3140 2012
[21] X Zhang and Y Han ldquoExistence and uniqueness of positivesolutions for higher order nonlocal fractional differential equa-tionsrdquo Applied Mathematics Letters vol 25 no 3 pp 555ndash5602012
[22] AM A El-Sayed andM Gaber ldquoTheAdomian decompositionmethod for solving partial differential equations of fractal orderin finite domainsrdquo Physics Letters A vol 359 no 3 pp 175ndash1822006
[23] H Jafari and V Daftardar-Gejji ldquoSolving a system of nonlinearfractional differential equations using Adomian decomposi-tionrdquo Journal of Computational and Applied Mathematics vol196 no 2 pp 644ndash651 2006
[24] K Diethelm and G Walz ldquoNumerical solution of fractionalorder differential equations by extrapolationrdquo Numerical Algo-rithms vol 16 no 3-4 pp 231ndash253 1997
[25] L Galeone and R Garrappa ldquoOn multistep methods for differ-ential equations of fractional orderrdquo Mediterranean Journal ofMathematics vol 3 no 3-4 pp 565ndash580 2006
[26] G Wang R P Agarwal and A Cabada ldquoExistence resultsand the monotone iterative technique for systems of nonlinearfractional differential equationsrdquo Applied Mathematics Lettersvol 25 no 6 pp 1019ndash1024 2012
[27] N J Ford and A C Simpson ldquoThe numerical solution of frac-tional differential equations speed versus accuracyrdquo NumericalAlgorithms vol 26 no 4 pp 333ndash346 2001
[28] V Lakshmikantham S Leela and J Vasundhara Theory ofFractional Dynamic Systems Cambridge Academic PublishersCambridge UK
[29] I Podlubny Fractional Differential Equations vol 198 ofMath-ematics in Science and Engineering Academic Press San DiegoCalif USA 1999
[30] S G Samko A A Kilbas and O I Marichev Fractionalintegrals and derivatives Gordon and Breach Science YverdonSwitzerland 1993
[31] C Li and Y Ma ldquoFractional dynamical system and its lineariza-tion theoremrdquo Nonlinear Dynamics vol 71 no 4 pp 621ndash6332013
[32] Z M Odibat and N T Shawagfeh ldquoGeneralized Taylorrsquosformulardquo Applied Mathematics and Computation vol 186 no1 pp 286ndash293 2007
[33] N Bellomo A Bellouquid J Nieto and J Soler ldquoMultiscalebiological tissue models and flux-limited chemotaxis for mul-ticellular growing systemsrdquoMathematical Models ampMethods inApplied Sciences vol 20 no 7 pp 1179ndash1207 2010
[34] AGokdogan A Yildirim andMMerdan ldquoSolving a fractionalordermodel ofHIV infection of CD+ T cellsrdquoMathematical andComputer Modelling vol 54 no 9-10 pp 2132ndash2138 2011
[35] D Kirschner and J C Panetta ldquoModeling immunotherapyof the tumormdashimmune interactionrdquo Journal of MathematicalBiology vol 37 no 3 pp 235ndash252 1998
[36] F A Rihan M Safan M A Abdeen and D Abdel RahmanldquoQualitative and computational analysis of a mathematicalmodel for tumor-immune interactionsrdquo Journal of AppliedMathematics vol 2012 Article ID 475720 19 pages 2012
[37] R Yafia ldquoHopf bifurcation in differential equations with delayfor tumor-immune system competition modelrdquo SIAM Journalon Applied Mathematics vol 67 no 6 pp 1693ndash1703 2007
[38] K A Pawelek S Liu F Pahlevani and L Rong ldquoA model ofHIV-1 infection with two time delays mathematical analysisand comparison with patient datardquo Mathematical Biosciencesvol 235 no 1 pp 98ndash109 2012
[39] A S Perelson D E Kirschner and R De Boer ldquoDynamics ofHIV infection of CD4+ T cellsrdquo Mathematical Biosciences vol114 no 1 pp 81ndash125 1993
[40] G A Bocharov and F A Rihan ldquoNumerical modelling inbiosciences using delay differential equationsrdquo Journal of Com-putational and Applied Mathematics vol 125 no 1-2 pp 183ndash199 2000
[41] R V Culshaw and S Ruan ldquoA delay-differential equationmodelof HIV infection of CD4+ T-cellsrdquo Mathematical Biosciencesvol 165 no 1 pp 27ndash39 2000
[42] R Anguelov and J M-S Lubuma ldquoNonstandard finite dif-ference method by nonlocal approximationrdquo Mathematics andComputers in Simulation vol 61 no 3-6 pp 465ndash475 2003
[43] C Li and F Zeng ldquoFinite difference methods for fractionaldifferential equationsrdquo International Journal of Bifurcation andChaos in Applied Sciences and Engineering vol 22 no 4 28pages 2012
[44] F A Rihan ldquoComputational methods for delay parabolicand time-fractional partial differential equationsrdquo NumericalMethods for Partial Differential Equations vol 26 no 6 pp1556ndash1571 2010
[45] Z Odibat and S Momani ldquoNumerical methods for nonlinearpartial differential equations of fractional orderrdquo Applied Math-ematical Modelling vol 32 no 1 pp 28ndash39 2008
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of