proofs10 sol
TRANSCRIPT
-
8/13/2019 Proofs10 Sol
1/4
Proofs Homework Set 10
MATH 217 WINTER 2011
Due March 23
PROBLEM 10.1. Suppose thatA andB aren n matrices that commute (that is, AB = BA)and suppose thatB hasndistinct eigenvalues.
(a) Show that ifBv= vthenBAv= Av.
Proof. This follows from the fact that AB = BA. Indeed,
BAv= ABv= A(v) =Av
since scalar multiplication commutes with matrix multiplication.
(b) Show that every eigenvector forB is also an eigenvector forA.
Proof. Suppose v is an eigenvector ofBwith eigenvalue . By part (a), we have BAv= Av.So eitherAv= 0orAvis also an eigenvector ofB with eigenvalue. SinceB hasndistincteigenvalues, they all have multiplicity 1 which means that all of the eigenspaces ofB are
one-dimensional (see Theorem 7(b) in Section 5.3). Since v and Av both lie in the one-dimensional eigenspace ofB corresponding to the eigenvalue , v and Av must be linearlydependent. Since v= 0, this means that Av = v for some scalar . Therefore, v is aneigenvector ofAcorresponding to the eigenvalue .
(c) Show that the matrixAis diagonalizable.
Proof. SinceB has n-distinct eigenvalues, we know thatB is diagonalizable by Theorem 6 ofSection 5.3. ThereforeB has nlinearly independent eigenvectors v1, . . . ,vnby Theorem 5 ofSection 5.3. By part (b), the vectors v1, . . . ,vn are also eigenvectors ofA. Therefore,A hasn linearly independent eigenvectors, which means that A is diagonalizable by Theorem 5 ofSection 5.3.
(d) Show that the matrixAB is diagonalizable.
Proof. By the solution of part (c) and Theorem 5 of Section 5.3, we have
A= P DP1, B= P EP1
-
8/13/2019 Proofs10 Sol
2/4
whereD and Eare diagonal matrices and
P =v1 vn
.
Note that we get the same matrix P forA and B since v1, . . . ,vn are eigenvectors of bothAandB . However, the eigenvalues corresponding to these eigenvectors may be different for A
andB so we get different diagonal matricesD and E.From this, we see that
AB= P DP1P EP1 =PDEP1,
which shows thatAB is diagonalizable sinceDEis a diagonal matrix.
PROBLEM10.2. The sequence of Lucas numbersis defined by the recurrence formula
L0= 2, L1= 1, Ln= Ln1+ Ln2for n 2.Thus, the sequence starts like this
2, 1, 3, 4, 7, 11, 18, 29, 47, 76, 123, . . .
In this problem, you will use linear algebra to find an explicit formula for Ln, the nth Lucasnumber.
LetA=
1 11 0
.
(a) Show thatA
LnLn1
=
Ln+1
Ln
for eachn 1.
Proof. We have
A Ln
Ln
1=
1 1
1 0
Ln
Ln
1=
Ln+ Ln1
Ln =
Ln+1
Ln
sinceLn+1= Ln+ Ln1by definition of the Lucas numbers.
(b) Use part (a) to prove by induction thatAn
12
=
Ln+1
Ln
holds for everyn 0.
Proof. Base case. First note thatA0 =Iby definition. Therefore, A0
12
=
12
=
L1L0
by
definition ofL0and L1.
Induction step. Suppose we know thatAn
12
=
Ln+1
Ln
; we want to show that An+1
12
=
Ln+2Ln+1
. Well, sinceAn+1 =AAn, we see that
An+1
12
= A
An
12
= A
Ln+1
Ln
by the Induction Hypothesis. But A
Ln+1
Ln
=
Ln+2Ln+1
by part (a), and so An+1
12
=
Ln+2Ln+1
as claimed.
-
8/13/2019 Proofs10 Sol
3/4
(c) Find the two eigenvalues for A. Call the positive one (this is the Greek letter phi) andverify that the negative one is equal to 1/.
Proof. The characteristic equation ofAis
det(A I) = det 1 11 = (1 )() 1 =2 1.The quadratic formula gives the two roots
1 +
5
2
152
.
The first root is clearly positive, and so = 12
(1 +
5) = 1.61803 . . . (This is the famous
Golden Ratio!) The second root is negative, 12
(15) = 0.61803 . . .The fact that the second root is
1/can be checked by multiplying the two numbers:
1 +
5
2
1 52
=1 5 + 5 5
4 =
44
= 1.
(d) Find eigenvectors corresponding to these two eigenvaluesand 1/.
Proof. For the first eigenvalue, we have
A I=
1 11
1 0 0
and thus we find the corresponding eigenvector
1
.
For the second eigenvalue 1/, we have
A +1
I=
1 + 1/ 1
1 1/
1 1/0 0
and thus we find the corresponding eigenvector
1/1
.
(e) Find an invertible matrixPsuch thatA= P DP1 whereDis a diagonal matrix.
Proof. By Theorem 5 of Section 5.3, we must haveA= P DP1 where
D=
00 1/
, P =
1/1 1
.
(f) First give an explicit formula forDn and use this to give an explicit formula forAn.
-
8/13/2019 Proofs10 Sol
4/4
Proof. Powers of a diagonal matrix are easy to compute explicitly:
Dn =
n 00 (1/)n
.
Then we use the formulaA
n
=P D
n
P
1
to explicitly computeA
n
as follows:
An =
1/1 1
n 00 (1/)n
1
+ 1/
1 1/1
which boils down to
An = 1
+ 1/
n+1 (1/)n+1 n (1/)n
n (1/)n n1 (1/)n1
.
(g) Using parts (b) and (f), give an explicit formula for thenth Lucas number!
Proof. Using the formulas from parts (b) and (f), we obtain
Ln+1
Ln
= An
12
=
1
+ 1/
n+1 (1/)n+1 n (1/)n
n (1/)n n1 (1/)n1
12
= 1
+ 1/
n+1 (1/)n+1 + 2n 2(1/)nn (1/)n + 2n1 2(1/)n1
=
n+1 + (1/)n+1
n + (1/)n
.
(For the last simplification step, it really helps to notice that + 1/ = 1 + 2/ = 2 1.)Therefore, we finally obtain the remarkable formula
Ln= n + (1/)n.