toward nnlo accuracy of parton distribution functions · toward nnlo accuracy of parton...
TRANSCRIPT
Toward NNLO accuracyof parton distribution functions
Pavel Nadolsky
Southern Methodist UniversityDallas, TX, U.S.A.
in collaboration withM. Guzzi, F. Olness,
J. Huston, H.-L. Lai, Z. Li, J. Pumplin, C.-P. Yuan (CTEQ)
September 23, 2011
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 1
NNLO PDF sets as the new norm
¥ PDFs with NNLO (two-loop)QCD corrections to DIS andDY processes are becomingthe standard. They are nowproduced by 6 groups.
¥ Our general-purpose setCT10 is obtained at NLO (PRD82,
074024 (2010)). The CT10.1 NNLO setundergoes pre-publicationtests.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 2
The agreement between NNLO PDF sets is not automatically
better than at NLO
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 3
The agreement between NNLO PDF sets is not automatically
better than at NLO
G. Watt, in PDF4LHC study, arXiv:1101.0536 )2Z
(MSα0.11 0.115 0.12 0.125 0.13
(p
b)
Hσ
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0.26
0.28
0.3
68% C.L. PDF
MSTW08
HERAPDF1.0
ABKM09
GJR08/JR09
= 180 GeVH
= 1.96 TeV) for MsH at the Tevatron (→NNLO gg
Open symbols: NLOClosed symbols: NNLO
SαOuter: PDF+Inner: PDF onlyVertical error bars
)2Z
(MSα0.11 0.115 0.12 0.125 0.13
(p
b)
Hσ
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0.26
0.28
0.3
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 3
χ2/Ndata points in various experiments (PRELIMINARY)PDF set Order All Combined BCDMS CDF, D0 D0 Run-2 Ae
ch,
expts. HERA-1 DIS Fp,d2 Run-2 1-jet pe
T > 25 GeV
CT10.1 1.11 1.17 1.10 1.33 3.72MSTW08 NLO 1.42 1.73 1.16 1.31 11.38
(1.28) (1.4) (1.17)NNPDF2.0 1.37 1.32 1.28 1.57 2.79
CT10.1 1.13 1.12 1.14 1.23 2.59MSTW08 1.34 1.36 1.15 1.38 9.84
NNPDF2.1 NNLO 1.57 1.36 1.30 1.51 5.45ABM’09 (5f) 1.65 1.4 1.49 2.63 23.78
HERA1.5 1.71 1.15 1.87 ? 5.4
Npoints 2798 579 590 182 12
Cross sections are computed using the CTEQ fitting code and αs, mc,mb values provided by each PDF set. Their agreement does notimmediately improve after going to NNLO.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 4
Origin of differences between PDF sets
¥ NNLO QCD terms (in all 6 PDF fits)
I Implementation of heavy-quark mass terms
¥ (N)LO electroweak contributions
¥ Selection of data: global analyses (CTEQ, MSTW, NNPDF) vs.restricted (“DIS-based”) analyses (ABM, GJR, HERAPDF)
¥ Statistical treatment: Monte-Carlo sampling vs. analyticalminimization of χ2; correlated systematic uncertainties;definitions of PDF uncertainties
¥ Initial PDF parametrizations: neural networks (NNPDF);2-5 parameters per flavor (other fits)
¥ Values of αs(MZ), mc, and mb and their treatment
¥ Differences in NLO codes used by PDF fits
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 5
This talk
Genuine NNLO accuracy requires an earnest effortto calibrate all components of the PDF fits
I will provide examples of related activities, focusing on
¥ Heavy-quark contributions to DIS at O(α2s)
M. Guzzi, P.N., H.-L. Lai, C.-P. Yuan, arXiv:1108.5112;
additional figures at http://bit.ly/SACOTNNLO11
¥ W charge asymmetry at the Tevatron1. H.-L. Lai, M. Guzzi, J. Huston, Z. Li, P. N., J. Pumplin, C.-P. Yuan, Phys.Rev. D82 (2010) 074024.
2. M. Guzzi, P. N., E. Berger, H.-L. Lai, F. Olness, C.-P. Yuan, arXiv:1101.0561 [hep-ph].
¥ Consistency of DGLAP picture at small x at HERA1. H.-L. Lai, M. Guzzi, J. Huston, Z. Li, P. N., J. Pumplin, C.-P. Yuan, Phys.Rev. D82 (2010) 074024.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 6
1. Heavy quarks in DIS and LHC electroweak cross sections
The latest PDF fits assume mc,b 6= 0 when evaluating Wilson coeffi-cients in DIS, e±p → e±X and e±p → νX
This is needed, in particular, to correctly predict W, Z productionrates at the LHC (Tung et al., hep-ph/0611254)
CMS-PAS-EWK-10-00518.5 19 19.5 20 20.5 21 21.5 22
ΣtotHpp®HW±®{ΝLXL HnbL
1.85
1.9
1.95
2
2.05
2.1
2.15
Σto
tHpp®HZ
0®{{-
LXLHn
bL
W± & Z cross sections at the LHC (14 TeV)
CTEQ6.6 (GM)
CTEQ6.1 (ZM)
NNLL-NLO ResBos
- arXiv:0802.0007
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 7
1. Heavy quarks in DIS and LHC electroweak cross sections
In pp → Z0X at√
s = 14 TeV:
σ(general-mass PDFs)σ(zero-mass PDFs)
≈ 1.05− 1.08;
to be compared with
KNNLO/NLO ≡ σ(α2s)/σ(αs) ≈ 1.02
CMS-PAS-EWK-10-00518.5 19 19.5 20 20.5 21 21.5 22
ΣtotHpp®HW±®{ΝLXL HnbL
1.85
1.9
1.95
2
2.05
2.1
2.15
Σto
tHpp®HZ
0®{{-
LXLHn
bL
W± & Z cross sections at the LHC (14 TeV)
CTEQ6.6 (GM)
CTEQ6.1 (ZM)
NNLL-NLO ResBos
- arXiv:0802.0007
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 7
Massive quark contributions to neutral-current DIS
Several heavy-quarkfactorization schemes
FFN, ACOT, BMSN, CSN,FONLL, TR’...
Extensive recent workTung et al., hep-ph/0611254; Thorne,hep-ph/0601245; Tung, Thorne, arXiv:0809.0714; P.N.,Tung, arXiv:0903.2667; Forte, Laenen, Nason,arXiv:1001.2312; J. Rojo et al., arXiv:1003.1241;Alekhin,Moch, arXiv:1011.5790;...
Is a consistent picture emerging? Willpersisting confusions (e.g., about theuniversality of heavy-quark PDFs) be
resolved?
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 8
Heavy-quark NC DIS at NLO: consistent ambiguityAt NLO, the charm mass mc and factorization scale µ of aretuned to best describe the DIS data in each scheme; but theresidual differences in the W and Z cross sections remain
à
à
à à
à
ò
ò
ò
ò
ò
æ
æ
æ æ
æ
ç
ç
ç ç
ç
Scale dependencelong dash: S-ACOT- Χ NLO
short dash: FFNS NLO Nf=3
à MSTW08-NLO
ò MSTW08-NLO- Χ
æ FONLL-A- Χ
ç FONLL-B- Χ
dotted:S-ACOT NLO
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
x
103x
0.5
F2
cHx
,QL
LH PDFs Q=2 GeV, mc=1.41 GeV
2009 Les Houches HQ benchmarkswith toy PDFs; default µ = Q
G. Watt, PDF4LHC mtg, 26.03.2010
W, Z cross sections;mc = 1.3 GeV in CTEQ6.6
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 9
NNLO: better agreement between the schemes,remaining conceptual differences
arXiv:1108.5112
¥ revisits the QCD factorizationtheorem for DIS with heavyquarks
¥ discusses a scheme(S-ACOT-χ) for a streamlined,algorithmic implementation ofNNLO massive contributions
Subtractions
c F(0) (1)h,h h,g
Structure Functions
ch,h(1)
A(1)h,g * c
h,h(1)
A(2)h,g ch,h
(0)* A
(2)* c(0)
h,hh,lPSPS
h,l(2)FFh,g
(2)
u,d,s,c
c(2)h,h
A(1)h,g c(0)
h,h*
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 10
S-ACOT-χ scheme: merging FFN and ZM
S-ACOT- Χ NNLO
FFNS Nf=3 NNLO
ZM NNLO
0.00
0.05
0.10
0.15
0.20
0.25
F2
cHx
,QL
x=0.01
10.5.2. 3.1.5 7.0.60.70.80.91.01.11.2
Q HGeVL
Rat
ioto
SA
CO
TΧ
S-ACOT-χ reducesto FFN at Q ≈ mc
and to ZM at Q À mc
Les Houches toy PDFs, evolvedat NNLO with threshold
matching terms
Cancellations betweensubtractions and other terms atQ ≈ mc and Q À mc; details inbackup slides
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 11
Are heavy quarks counted as active partons?Fixed Flavor Number schememc 6= 0 for all µ2 = Q2 ≥ m2
c
PDF"hard" part
massive quark
Zero-Mass Variable FlavorNumber scheme
mc = 0 for all Q2 ≥ m2c
massless quark
fc/p(ξ, µ2) ∼∞X
n=1
αkS(µ)
nX
k=0
cnk(ξ) lnk
„µ2
m2c
«
Shown for up to 4 flavors for simplicity
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 12
General-Mass Variable Flavor Number schemes
(Aivasis, Collins, Olness, Tung; Buza et al.; Cacciari, Greco, Nason;
Chuvakin et al.; Kniehl et al.; Thorne, Roberts; Forte, Laenen, Nason; ...)
For the ACOT GM scheme,factorization of DIS cross sections isproved to all orders of αs (Collins, 1998)
Theorem
F2(x,Q, mc) = σ0
∑a=g,q,q
∫dξ
ξCa
(x
ξ,
Q
µF,mc
Q
)fa/p(ξ,
µF
mc)+O
(ΛQCD
Q
)
¥ Ca is a Wilson coefficient with an incoming parton a = g, u(−)
, ..., c(−)
¥ fa/p is a PDF for Nf flavors
¥ Nf = 3 for µ < µ(4)switch ≈ mc; Nf = 4 for µ ≥ µ
(4)switch
¥ limQ→∞ Ca exists; no terms O(mc/Q) in the remainderPavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 13
General-Mass Variable Flavor Number schemes
(Aivasis, Collins, Olness, Tung; Buza et al.; Cacciari, Greco, Nason;
Chuvakin et al.; Kniehl et al.; Thorne, Roberts; Forte, Laenen, Nason; ...)
For the ACOT GM scheme,factorization of DIS cross sections isproved to all orders of αs (Collins, 1998)
Schemes of the ACOT type do not use...
¥ PDFs for several Nf values in the same µ range
¥ smoothness conditions/damping factors at Q → mc
¥ constant terms from higher orders in αs to ensure continuityat the threshold
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 13
Components of inclusive F2,L(x,Q)Structure of S-ACOT-χ NNLO expressions is reminiscent of the ZM scheme(e.g., in Moch, Vermaseren, Vogt, 2005)
Components of inclusive F2,L(x,Q2) are classified according tothe quark couplings to the photon
F =
Nl∑
l=1
Fl + Fh (1)
Fl = e2l
∑a
[Cl,a ⊗ fa/p
](x,Q), Fh = e2
h
∑a
[Ch,a ⊗ fa/p
](x,Q). (2)
e l he
AtO(α2
s):
F(2)h = e2
h
{cNS,(2)h,h ⊗ (fh/p + fh/p) + C
(2)h,l ⊗ Σ + C
(2)h,g ⊗ fg/p
}
F(2)l = e2
l
{C
NS,(2)l,l ⊗ (fl/p + fl/p) + cPS,(2) ⊗ Σ + c
(2)l,g ⊗ fg/p
}. (3)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 14
Components of inclusive F2,L(x,Q)
¥ Lower case c(2)a,b, f
(k)a,b → ZM expressions
Zijlstra and Van Neerven PLB272 (1991), NPB383 (1992)S. Moch, J.A.M. Vermaseren and A. Vogt, NPB724 (2005)
¥ Upper case C(2)a,b , F
(k)a,b A
(k)a,b → coeff. functions, structure
functions and subtractions with mc 6= 0,Buza et al., NPB 472 (1996); EPJC1 (1998);Riemersma, et al. PLB 347 (1995); Laenen et al. NPB392 (1993)
¥ All building blocks are available from literature
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 14
Components of inclusive F2,L(x,Q)The separation into Fl and Fh (according to the quark’s electriccharge e2
i ) is valid at all Q
The “light-quark” Fl contains some subgraphs with heavy-quarklines, denoted by “Gl,l,heavy”.
The “heavy-quark” Fh 6= F c2 :
F c2 = Fh + (Gl,l,heavy)real,
where Gi,j = C(2)i,j , F
(2)i,j , and A
(2)i,j
l
h
(a) (b)
l
h
(c) (d)
l
hh
l
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 14
Three versions of the ACOT schemeScheme Lowest-order graphs Variables in Hc
Full ACOTAivasis et. al.,
hep-ph/9312319
Proof in Collins,
hep-ph/9806259
+ −P1m;n=1�nSvnm lnm(Q2=M2)m!mc 6= 0
ζ =
x2
(1 +
√1 + 4m2
c
Q2
)
SimplifiedACOT
Proof in Collins,hep-ph/9806259;
Kramer, Olness, Soper,
hep-ph/0003035
+ − mc = 0ζ = x
S-ACOT-χTung, Kretzer, Schmidt,
hep-ph/0110247;
proof in Guzzi et al.,
arXiv:1108.5112
+ − mc = 0ζ = x
(1 +
4m2c
Q2
)≡ χ
¥ Ha with incoming light partons are unambiguous¥ Hc with incoming c quarks differ by terms of order (m2
c/Q2)p, p > 0Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 15
Three versions of the ACOT schemeScheme Lowest-order graphs Variables in Hc
Full ACOTAivasis et. al.,
hep-ph/9312319
Proof in Collins,
hep-ph/9806259
+ −P1m;n=1�nSvnm lnm(Q2=M2)m!mc 6= 0
ζ =
x2
(1 +
√1 + 4m2
c
Q2
)
SimplifiedACOT
Proof in Collins,hep-ph/9806259;
Kramer, Olness, Soper,
hep-ph/0003035
+ − mc = 0ζ = x
S-ACOT-χTung, Kretzer, Schmidt,
hep-ph/0110247;
proof in Guzzi et al.,
arXiv:1108.5112
+ − mc = 0ζ = x
(1 +
4m2c
Q2
)≡ χ
¥ Ha with incoming light partons are unambiguous¥ Hc with incoming c quarks differ by terms of order (m2
c/Q2)p, p > 0Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 17
Energy conservation near the thresholdThreshold rate suppression resulting from energy conservation isthe most important mass effect at Q ≈ mc
(Tung, Kretzer, Schmidt, hep-ph/0110247; Tung, Thorne, arXiv:0809.0714; P.N., Tung, arXiv:0903.2667)
In DIS, a cc pair is produced only at par-tonic energies W satisfying
W 2 = (ξp + q)2 = Q2
(ξ
x− 1
)≥ 4m2
c
⇒χ ≤ ξ ≤ 1, where χ ≡ x
(1 +
4m2c
Q2
)≥ x.
/ Collinear approximation for cc pro-duction may allow to integrate over
x ≤ ξ ≤ 1,
in violation of energy conservation
eξpq
c
cp
Threshold
suppression
x=0.05 µ2=Q
2+4 m
2
F2c
Q2 / GeV
2
χ=x(1+4 m2 / Q
2 )
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
1 10 102
103
NLO 3-flv
LO 3-flv : GF(1)
Naive LO 4-flv : c
(x)
LO
4-f
lv A
CO
T(c)
: c(c)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 18
Energy conservation near the thresholdThreshold rate suppression resulting from energy conservation isthe most important mass effect at Q ≈ mc
(Tung, Kretzer, Schmidt, hep-ph/0110247; Tung, Thorne, arXiv:0809.0714; P.N., Tung, arXiv:0903.2667)
, The S-ACOT-χ scheme allows oneto include the energy conservation re-quirement in the QCD factorization the-orem (arXiv:1108.5112)
, This is achieved by rescaling, whichrestores the correct kinematics of HQproduction at the threshold without aposteriori constraints or damping fac-tors imposed by other schemes
eξpq
c
cp
Threshold
suppression
x=0.05 µ2=Q
2+4 m
2
F2c
Q2 / GeV
2
χ=x(1+4 m2 / Q
2 )
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
1 10 102
103
NLO 3-flv
LO 3-flv : GF(1)
Naive LO 4-flv : c
(x)
LO
4-f
lv A
CO
T(c)
: c(c)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 18
Rescaling at the lowest order
+ −
c(ζ) +αs
4π
Z 1
χ
dξ
ξg(ξ)C
(1)h,g
„χ
ξ
«− αs
4π
Z 1
ζ
dξ
ξg(ξ)A
(1)h,g
„ζ
ξ
«
®
©
ªζ takes place of x
in terms 1 and 3
¥ Term 2 (γ∗g fusion) is unambiguous
¥ Terms 1 and 3 are essentiallyZ 1
0
dξ
ξδ
„1− ζ
ξ
«c(ξ)− αs
4π
Z 1
0
dξ
ξg(ξ)A
(1)h,g
„ζ
ξ
«θ
„1− ζ
ξ
«.
Rescaling ζ → κζ, ξ → κ−1ξ changes the ξ range in theWilson coefficients δ(1− ζ
ξ ) and A(1)h,g
(ζξ
)θ(1− ζ
ξ
), but does
not change their magnitude
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 19
Rescaling at the lowest order
+ −
c(ζ) +αs
4π
Z 1
χ
dξ
ξg(ξ)C
(1)h,g
„χ
ξ
«− αs
4π
Z 1
ζ
dξ
ξg(ξ)A
(1)h,g
„ζ
ξ
«
®
©
ªζ takes place of x
in terms 1 and 3
Q2 = 10 GeV2
¥ Red curve: g(ξ)C(1)h,g(χ/ξ) at χ ≤ ξ ≤ 1
¥ Green: ζ = x; κ = 1
I g(ξ)A(1)h,g(x/ξ) 6= 0 at ξ < χ
I its integral cancels poorly with c(x)
¥ Blue: ζ = χ; κ = 1 + 4m2c/Q2
Ig(ξ)A(1)h,g(χ/ξ) = 0 at ξ < χ
I its integral cancels better with c(χ)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 19
Rescaling at the lowest order
+ −
c(ζ) +αs
4π
Z 1
χ
dξ
ξg(ξ)C
(1)h,g
„χ
ξ
«− αs
4π
Z 1
ζ
dξ
ξg(ξ)A
(1)h,g
„ζ
ξ
«
®
©
ªζ takes place of x
in terms 1 and 3
Q2 = 100 GeV2
χ ≈ x
g(ξ)A(1)h,g(ζ/ξ) approximates the logarithmic
growth in g(ξ)C(1)h,g(χ/ξ)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 19
Rescaling at the lowest order
+ −
c(ζ) +αs
4π
Z 1
χ
dξ
ξg(ξ)C
(1)h,g
„χ
ξ
«− αs
4π
Z 1
ζ
dξ
ξg(ξ)A
(1)h,g
„ζ
ξ
«
®
©
ªζ takes place of x
in terms 1 and 3
Q2 = 10000 GeV2
χ ≈ x
g(ξ)A(1)h,g(ζ/ξ) approximates the logarithmic
growth in g(ξ)C(1)h,g(χ/ξ)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 19
Rescaling to all orders of αs and the factorization theorem
Z projectionk
k
pa = g, q, q
q qp
~kT
Ta(k, p)Ta(k, p) Ha(q,k)Ha(q, k) ⊗
¥ Rescaling is introduced in the definition of the hard subgraphHc(q, k) with an incoming c quark in γ∗(q) + c(k) → X. Hard graphsHg, Hq with incoming light partons and target graphs Ta(k, p) are notaffected
¥ In Hc(q, k), the momentum of the incoming c quark is approxi-mated by bk =
“ξp+, 0−,~0T
”. k and Hc(q, k) are invariant under trans-
formation p+ → p+/κ, ξ → ξ κ. The physical ξ range is obtained forκ = 1 + 4m2
c/Q2.
¥ This transformation shows that the S-ACOT-χ scheme is valid toall orders on the same grounds as the S-ACOT scheme.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 20
Rescaling to all orders of αs and the factorization theorem
Z projectionk
k
pa = g, q, q
q qp
~kT
Ta(k, p)Ta(k, p) Ha(q,k)Ha(q, k) ⊗
F (x, Q) =X
a=g,u,d,...,c
Zdξ
ξCa
„x
ξ,Q
µ,mc
Q
«fa/p(ξ, µ)
¥ Wilson coefficients with initial heavy quarks are
Cc
„x
ξ,Q
µ,mc
Q
«≈ Cc
„χ
ξ,Q
µ, mc = 0
«θ(χ ≤ ξ ≤ 1)
where χ ≡ x
„1 +
4m2c
Q2
«.
¥ The target (PDF) subgraphs Ta are given by the same universaloperator matrix elements in all ACOT schemes
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 20
NNLO results for F(c)2 (x,Q2)
At NLO:
à
à
à à
à
ò
ò
ò
ò
ò
æ
æ
æ æ
æ
ç
ç
ç ç
ç
Scale dependencelong dash: S-ACOT- Χ NLO
short dash: FFNS NLO Nf=3
à MSTW08-NLO
ò MSTW08-NLO- Χ
æ FONLL-A- Χ
ç FONLL-B- Χ
dotted:S-ACOT NLO
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
x
103x
0.5
F2
cHx
,QL
LH PDFs Q=2 GeV, mc=1.41 GeV
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 21
NNLO results for F(c)2 (x,Q2)
At NNLO and Q ≈ mc:
S-ACOT-χ (Nf = 4) ≈ FFN (Nf = 3)
without tuning
¥ S-ACOT is numerically closeto other NNLO schemes
¥ NNLO expressions are closeto the FONLL-C scheme(Forte, Laenen, Nason, arXiv:1001.2312).
ò
ò
ò
ò
ò
æ
æ
æ æ
æ
Scale dependencelong dash: S-ACOT- Χ NLO
solid: S-ACOT- Χ NNLO
short dash: FFNS NNLO Nf=3
ò MSTW08-NNLO- Χ
æ FONLL-C- Χ
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
x
103x
0.5
F2
cHx
,QL
LH PDFs Q=2 GeV, mc=1.41 GeV
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 21
NNLO results for F(c)2 (x,Q2)
At NNLO and Q ≈ mc:
S-ACOT-χ (Nf = 4) ≈ FFN (Nf = 3)
without tuning
¥ S-ACOT is numerically closeto other NNLO schemes
¥ NNLO expressions are closeto the FONLL-C scheme(Forte, Laenen, Nason, arXiv:1001.2312).
scale+rescaling dependenceblue band: NLO
green band: NNLO
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20
1
2
3
4
5
x
103x
0.5
F2
cHx
,QL
LH PDFs Q=2 GeV S-ACOT
¥ Even without rescaling (a wrong choice!), NNLO cross sectionsare much closer to FFN at Q ≈ mc than at NLO
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 21
Main features of the S-ACOT-χ scheme¥ It is proved to all orders by the QCD factorization theorem for
DIS (Collins, 1998)
¥ It is relatively simple
I One value of Nf (and one PDF set) in each Q range
I sets mh = 0 in ME with incoming h = c or b
I matching to FFN is implemented as a part of the QCDfactorization theorem
¥ Universal PDFs¥ It reduces to the ZM MS scheme at Q2 À m2
Q, withoutadditional renormalization
¥ It reduces to the FFN scheme at Q2 ≈ m2Q
I has reduced dependence on tunable parameters at NNLO
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 22
CT10/CT10W distributions, W charge asymmetry,HERA DIS at small x
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 23
CT10 parton distribution functions (PRD82, 074024 (2010))
New experimental data, statistical methods, andparametrization forms
¥ combined HERA-1 data are included
¥ 53 CT10 and 53 CT10W eigenvector sets for αs(MZ) = 0.118
¥ 4 CT10AS PDFs for αs(MZ) = 0.116− 0.120sufficient for computing the PDF+αs uncertainty, as explained inPRD82,054021 (2010)
¥ CT10/CT10W PDFs with 3 and 4 active flavors
¥ extended discussion ofthe Tevatron Run-2 W asymmetry data
¥ search for deviations from NLO DGLAP evolutionin the small-x HERA data
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 24
The puzzle of the W lepton asymmetry
¥ Rapidity asymmetry A`(y`) of charged lepton ` = e or µ inW boson decay provides important constraints ond(x,Q)/u(x,Q) at x > 0.1
A`(y`) =dσ(pp → (W+ → `+ν`)X)/dy` − dσ(pp → (W− → `−ν`)X)/dy`
dσ(pp → (W+ → `+ν`)X)/dy` + dσ(pp → (W− → `−ν`)X)/dy`.
¥ A`(y`) is related to the boson-level asymmetry,
AW (yW ) =dσ(pp → W+X)/dyW − dσ(pp → W−X)/dyW
dσ(pp → W+X)/dyW + dσ(pp → W−X)/dyW,
smeared by W± decay effectsBerger, Halzen, Kim, Willenbrock, PRD 40, 83 (1989); Martin, Roberts, Stirling, 1989
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 25
The puzzle of the CDF/D0 W lepton asymmetry
¥ CT10W set reasonably agrees with 3 pT` bins of Ae(ye) andone bin of Aµ(yµ) from D0 Run-2 (2008); shows tension withNMC, BCDMS F p,d
2 (x,Q)
¥ NNPDF 2.0 (arXiv:1012.0836) agrees with Aµ(yµ), disagrees with twopTe bins of Ae(ye).
¥ CT10, many other PDFs fail.
Agreement of Source orPQCD with D0 Ae(ye) χ2/npt comments
CTEQ6.6, NLO 191/36=5.5 Our study;
CT10W, NLO 78/36=2.2 Resbos, NNLL-NLO
With Aµ(yµ): 88/47=1.9ABKM’09, NNLO 540/24=22.5 Catani, Ferrera, Grazzini,
MSTW’08, NNLO 205/24=8.6 JHEP 05, 006 (2010)
JR09VF, NNLO 113/24=4.7
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 26
Why difficulties with fitting A`(y`)?
1. A`(y`) is very sensitive to the average slope sdu ofd(x, MW )/u(x, MW )
A`(y`) ∼ A`(yW )|LO ∝1
x1 − x2
[d(x1)
u(x1)− d(x2)
u(x2)
]; x1,2 =
Q√se±yW
Berger, Halzen, Kim, Willenbrock, PRD 40, 83 (1989); Martin, Stirling, Roberts, MPLA 4, 1135 (1989); PRD D50, 6734 (1994);Lai et al., PRD 51, 4763 (1995)
2. Constraints on sdu by fixed-target F d2 (x,Q)/F p
2 (x,Q) areaffected by nuclear and higher-twist effectsAccardi, Christy, Keppel, Monaghan, Melnitchouk, Morfin, Owens, PRD 81, 034016 (2010)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 27
Challenges with fitting A`(y`)
Small changes in sdu causesignificant variations in A`
Lai et al., PRD 51, 4763 (1995)
Alternative constraints on d/uby F d
2 (x,Q)/F p2 (x,Q) from
fixed-target DIS are affectedby nuclear and higher-twisteffectsAccardi, Christy, Keppel, Monaghan, Melnitchouk,Morfin, Owens, PRD 81, 034016 (2010)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 28
d(x,Q)/u(x,Q) at Q = 85 GeV
η0 0.5 1 1.5 2 2.5 3
)ηA
(
-0.8
-0.6
-0.4
-0.2
0
0.2NNPDF20
CT10
MSTW08
>25GeV)T
D0el(E
R. Ball et al., arXiv:1012.0836
R. Ball et al., arXiv:1012.0836 H.-L. Lai et al., arXiv:1007.2241
Solid band: CTEQ6.6 uncertaintyHatched band: CT10W uncertainty
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.2 0.5 0.70.7
0.8
0.9
1.0
1.1
1.2
1.3
x
Rat
ioto
CT
EQ
6.6M
d/u at µ = 85 GeV
H.-L. Lai et al., arXiv:1007.2241
¥ D0 Run-2 A` data distinguishes between PDF models, reducesthe PDF uncertainty
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 29
Why difficulties with fitting A`(y`)?
3. Existing parametrizations underestimate the PDF uncertaintyon d/u
PDFs based on Chebyshovpolynomials improveagreement with D0 Run-2 Ae,but are outside of currentCTEQ/MSTW bands (Pumplin)
This ambiguity is reduced byA`(y`) at the LHC, whichconstrains d/u and d/u atx ∼ 0.01.
Band: CT10 uncertainty
Long dash: Chebyshov, dbar/ubar->1 at x->0
Short dash: Chebyshov, free dbar/ubar at x->0
Solid: MSTW’2008NLO
PRELIMINARY
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 30
CT10(W) vs. A` at the LHC
|µη|0 0.5 1 1.5 2
µA
0.15
0.2
0.25
0.3
0.35
|µη|0 0.5 1 1.5 2
µA
0.15
0.2
0.25
0.3
0.35=7 TeV)sData 2010 (
MC@NLO, CTEQ 6.6MC@NLO, HERA 1.0MC@NLO, MSTW 2008
-1 L dt = 31 pb∫
νµ →W
ATLAS
CT10(W) agrees well with theLHC A`; some differencesbetween NLO and NNLL+NLO
|η|0 0.5 1 1.5 2 2.5 3
W c
har
ge
asym
met
ry
0.1
0.15
0.2
0.25
0.3
0.35 CT10
CT10W
CMS electron
CMS muon
NNLL−NLO+K, ResBos
PRELIMINARY
Zhao Li, 2011
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 31
CT10(W) vs. A`: LHC-B
|η|1.5 2 2.5 3 3.5 4 4.5
W c
harg
e as
ymm
etry
-0.6
-0.4
-0.2
0
0.2
0.4
CT10
CT10W
LHCBNNLL−NLO+K, ResBos
PRELIMINARY
Zhao Li, 2011
LHC-B marginally prefers CT10W
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 32
Why difficulties with fitting A`(y`)?
4. Experimental A` with lepton pT` cuts is sensitive to dσ/dqT of Wboson at transverse momentum qT → 0.
¥ Fixed-order (N)NLO calculations (DYNNLO, FEWZ, MCFM,...)predict a wrong shape of dσ/dqT at qT → 0.
¥ Small-qT resummation correctly predicts dσ/dqT in this limit.
¥ CT10(W) PDFs are fitted using a NNLL-NLO+K resummedprediction for A` (ResBos); must not be used with fixed-orderpredictions for A`.
For example:
χ2(CT10W+ResBos) = 1.9Npt (us);
χ2(CT10W+DYNNLO) = 8.4Npt (NNPDF)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 33
Radiative contributions in ResBos +comparison to the Tevatron Z qT distribution
¥ Resummed dσ`pp → (V → ` ¯′)X
´/(d3p`d3p ¯′ ) for
V = γ∗, W, Z, with decay and EWinterference effects
¥ QT ∼ Q: [O(αs) with full decay] ×[average O(α2
s) correction (Arnold, Reno)]
¥ QT ¿ Q: A(3), B(2), C(1)
I no H(2) correction – a ≈ constanterror of ±3% in the normalization ofXsecs
¥ mass dependence in c, b scatteringcross sections (S. Berge, P. N., F. I. Olness, hep-ph/0509023)
¥ a nonperturbative function fromKonychev, P. N., hep-ph/0506225
(1/
GeV
)
T/d
pZσ
d× Zσ
1/
-610
-510
-410
-310
-210
-110
-1, L=0.97 fb∅DNNLO pQCD + corr.
PYTHIA Perugia 6
< 115 GeVµµ65 < M > 15 GeV
T| < 1.7, muon pηmuon |
∅D
(1/
GeV
)
T/d
pZσ
d× Zσ
1/
-610
-510
-410
-310
-210
-110
Rat
io to
PY
TH
IA P
erug
ia 6
1
-1, L=0.97 fb∅D
PYTHIA scale unc.
HERWIG+JIMMY
PYTHIA Tune D62
0.7
0.5
Rat
io to
PY
TH
IA P
erug
ia 6
1
(GeV)ZT
p1 10 210
Rat
io to
PY
TH
IA P
erug
ia 6
1
-1, L=0.97 fb∅DNNLO pQCD
Scale & PDF unc.
RESBOSNLO pQCD
Scale & PDF unc.
PYTHIA scale unc.
2
0.7
0.5
(GeV)ZT
p1 10 210
Rat
io to
PY
TH
IA P
erug
ia 6
1
(GeV)ZT
p1 10 210
Rat
io to
PY
TH
IA P
erug
ia 6
1
-1, L=0.97 fb∅DSHERPAPYTHIA scale unc.
ALP+HERWIGALP+PY Tune D6ALP+PY Perugia 6
2
0.7
0.5
(GeV)ZT
p1 10 210
Rat
io to
PY
TH
IA P
erug
ia 6
1
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 34
Charge asymmetry in peT bins (CDF Run-2, 207 pb−1)
Without the peT cut (FEWZ):
Anastasiou et al., 2003
y
y of W boson
With pTe cuts imposed, Ach(ye) is sensitive to small-QT resummation
arXiv:1101.05610.5 1 1.5 2 2.5 3
y{
-0.6
-0.4
-0.2
0
0.2
Cha
rge
asym
met
ry
25 < pT{ < 35 GeV, CTEQ65
Resummed
NLOLO
0.5 1 1.5 2 2.5 3y{
-0.4
-0.2
0
0.2
0.4
Cha
rge
asym
met
ry
35 < pT{ < 45 GeV, CTEQ65
NNLL�NLO
NLO
LO
Balazs, Yuan, 1997; PN, 2007, unpublished; arXiv:1101.0561
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 35
Acut fits to combined HERA data
x-510 -410 -310 -210 -110 1
]2 [
GeV
2 T /
p2
/ M
2Q
1
10
210
310
410
510
610 NMC-pdNMCSLACBCDMSHERAI-AVCHORUSFLH108NTVDMNZEUS-H2DYE886CDFWASYCDFZRAPD0ZRAPCDFR2KT
= 0.5cutA = 1.0cutA = 1.5cutA = 3.0cutA = 6.0cutA
NNPDF2.0 dataset
Ags ≡ Q2x0.3 < Acut
Fitting procedure:¥ Include only DIS data above an Acut line
¥ Compare the resulting PDFs with DIS data below the Acut
line, in a region that is “connected” by DGLAP evolutionPavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 36
CT10: Acut fits to DIS data at Q > Q0 = 2 GeV>2
<d
0.5
1
1.5
2
2.5
3 = 1.5cutFit with A
Fit without cuts < 1.0 cutA
< 1.5cut1.0 < A
< 3.0cut1.5 < A
< 6.0 cut3.0 < A
> 6.0cutA
Caola, Forte, Rojo, arXiv:1007.5405v2
Q > 1.41 GeV
<1.0 1.0-1.5 1.5-3.0 3.0-6.0 >6.0
ææ
æ
ææ
à
à
à
àà
ì
ì
ì
ìì
CT10 fit
Fit 1 with Acut=1.5Fit 2 with Acut=1.5
Q > 2 GeV
N pts= 36 18 30 27 468Ags
ranges
0.5
1.0
1.5
2.0
2.5
3.0Χdata
2�Npts
Motivation
Search for deviations from DGLAP evolution at smallest x and Q
¥ Follow the procedure proposed by NNPDF (Caola, Forte, Rojo, arXiv:1007.5405)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 37
CT10: Acut fits to DIS data at Q > Q0 = 2 GeV>2
<d
0.5
1
1.5
2
2.5
3 = 1.5cutFit with A
Fit without cuts < 1.0 cutA
< 1.5cut1.0 < A
< 3.0cut1.5 < A
< 6.0 cut3.0 < A
> 6.0cutA
Caola, Forte, Rojo, arXiv:1007.5405v2
Q > 1.41 GeV
<1.0 1.0-1.5 1.5-3.0 3.0-6.0 >6.0
ææ
æ
ææ
à
à
à
àà
ì
ì
ì
ìì
CT10 fit
Fit 1 with Acut=1.5Fit 2 with Acut=1.5
Q > 2 GeV
N pts= 36 18 30 27 468Ags
ranges
0.5
1.0
1.5
2.0
2.5
3.0Χdata
2�Npts
CT10
Two CT10-like fits to data at Ags > 1.5, with differentparametrizations of g(x,Q)
χ2i =
(Shifted Data− Theory)2
σ2uncor
Large syst. shifts at Ags < 1.0, in apattern that could mimic aslower Q2 evolution
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 37
CT10: Acut fits to DIS data at Q > Q0 = 2 GeV>2
<d
0.5
1
1.5
2
2.5
3 = 1.5cutFit with A
Fit without cuts < 1.0 cutA
< 1.5cut1.0 < A
< 3.0cut1.5 < A
< 6.0 cut3.0 < A
> 6.0cutA
Caola, Forte, Rojo, arXiv:1007.5405v2
Q > 1.41 GeV
<1.0 1.0-1.5 1.5-3.0 3.0-6.0 >6.0
ææ
æ
ææ
à
à
à
àà
ì
ì
ì
ìì
CT10 fit
Fit 1 with Acut=1.5Fit 2 with Acut=1.5
Q > 2 GeV
N pts= 36 18 30 27 468Ags
ranges
0.5
1.0
1.5
2.0
2.5
3.0Χdata
2�Npts
CT10, cont.
δχ2 ∼ 0 at Ags > 1.0[no difference]
δχ2 = 0− 1.5 at Ags < 1.0,with large uncertainty
⇒ Disagreement with the “DGLAP-connected” data atAgs < Acut is not supported by the CT10 fit
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 37
Conclusions¥ Great progress in understanding of heavy-quark
contributions to DIS.
I S-ACOT-χ recipe-like formulas for implementing NNLO
I Energy conservation in DIS is realized as a part of the QCDfactorization theorem. Would it be interesting to try to extendto pp processes.
I It leads to rescaling of Wilson coefficient functions withincoming heavy quarks. The PDFs are given by universaloperator matrix elements.
¥ Progress in understanding of NNLO contributions, newTevatron and LHC data sets, PDF parametrization issues
¥ arXiv:1101.0561: synopsis of recent CTEQ publications
I CT10W fit to Run-2 W charge asymmetry;PDFs for leading-order showering programs; constraints oncolor-octet fermions
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 38
Conclusion
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 39
Backup slides1. Details on S-ACOT-χ scheme at NNLO
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 40
S-ACOT input parameters
At Q ≈ mc, F c2 depends significantly on
1. Charm mass: mc = 1.3 GeV in CT10
2. Factorization scale: µ =√
Q2 + κm2c ; κ = 1 in CT10
3. Rescaling variable ζ(λ) for matching in γ∗c channels(Tung et al., hep-ph/0110247; Nadolsky, Tung, PRD79, 113014 (2009))
Fi(x,Q2) =∑
a,b
∫ 1
ζ
dξ
ξfa(ξ, µ) Ca
b,λ
(ζ
ξ,Q
µ,mi
µ
)
x = ζ/(
1 + ζλ · (4m2c)/Q2
), with 0 ≤ λ . 1
CT10 usesζ(0) ≡ χ ≡ x
(1 + 4m2
c/Q2),
motivated by momentum conservation
1+ �����������Mf
2
Q2 Ζ=ΧACOT HΛ=0L
Ζ=x HΛ®¥L
Λ=0.1
Λ=0.2
Λ=1 Phy
sica
lthr
esho
ld:W=
Mf;Ζ=
1
10-4 0.001 0.01 0.1 1
1
x
Res
calin
gfa
ctor�x
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 41
Classes of Feynman diagrams I
+
+
+ + +
NLO Subtraction
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 42
Classes of Feynman Diagrams II
− − −
+ + +...
−
NNLO Subtractions
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 43
Cancellations between Feynman diagrams
Validity of the S-ACOT calculation was verified by checking forcertain cancellations at Q ≈ mc and Q À mc
¥ Q ≈ mc:
D(2)C1 ¿ D
(2)C0 ¿ D
(1)C0 ≤ F c
2 (x,Q)
¥ Q À mc:
D(2)g ¿ D(1)
g < F c2 (x,Q)
These cancellations are indeed observed in our results
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 44
NNLO: Cancellations at Q2 ≈ m2c
D(1)
C0
D(1)C0 = C
(0)Q ⊗ c− asC
(0)Q ⊗A
(1)Qg ⊗ g; as =
αs
(4π)(1)
D(2)
C0
D(2)C0 = D
(1)C0 − a2
sC(0)Q ⊗A
(2),SQg ⊗ g − a2
sC(0)Q ⊗A
(2),PSQΣ ⊗ Σ (2)
D (2)
C1
D(2)C1 = C
(1)Q ⊗ c− a2
s C(1)Q ⊗A
(1)Qg ⊗ g (3)
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 45
NNLO: Cancellations at Q2 ≈ m2c
FFNS nf=3 nnl
LO OHΑs0L
DC0H1L
DC0H2L
DC1H2L
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20
1
2
3
4
x
103x
0.5
F2cHx
,QL
S−ACOT (no rescaling), Q=2 GeV
FFNS nf=3 nnl
LO OHΑs0L
DC0H1L
DC0H2L
DC1H2L
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.20
1
2
3
4
x
103x
0.5
F2cHx
,QL
S−ACOT−chi, Q = 2 GeV
D(2)C1 ¿ D
(2)C0 ¿ D
(1)C0 ≤ FFN at NNLO both for ζ = x and ζ = χ.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 46
NNLO: Cancellations at Q À mc
D(1)g
D(1)g ≡ C(1)
g = as
“F (1)
g ⊗ g − C(0)Q ⊗A
(1),SQg ⊗ g
”(4)
+ +
(2)gD
D(2)g = D(1)
g + a2s
hF (2)
g ⊗ g + F(2)Σ ⊗ Σ− C
(1)Q ⊗A
(1),SQg ⊗ g
−C(0)Q ⊗A
(2),SQg ⊗ g − C
(0)Q ⊗A
(2),PSQΣ ⊗ Σ
i(5)
D(1)g is of order of α2
s while D(2)g is of order of α3
s.Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 47
F c2 at NNLO: Cancellations at Q = 10 GeV
FFNS nf=3 nnl
DgH1L
DgH2L
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.2
0
5
10
15
x
103x
0.5
F2cHx
,QL
S−ACOT (no rescaling), Q = 10 GeV
FFNS nf=3 nnl
DgH1L
DgH2L
10-5 10-4 10-3 0.01 0.02 0.05 0.1 0.2
0
5
10
15
x
103x
0.5
F2cHx
,QL
S−ACOT−chi, Q = 10 GeV
D(2)g ¿ D
(1)g < FFN at NNLO < ACOT
log Q2
m2c
terms in FFN are cancelled well by subtractions.
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 48
Comparison of (N)NLO PDF sets with data in the CT10.1 fit
¥ χ2 are computed at NNLO, using the LHAPDF 5.8.6 interfaceand CTEQ fitting code (very naively!)
¥ Whenever possible, adjust settings to reproduce assumptionsby other groups
I Use αs(Mz), mc,b values suggested by each PDF set
I approximate the GM scheme in DIS if possible
¥ Correlated systematic errors are included according to theCTEQ method
χ2 =∑
e={expt.}
Npt∑
k=1
1
s2k
(Dk − Tk −
Nλ∑
α=1
λαβkα
)2
+
Nλ∑
α=1
λ2α
N Dk and Tk are data and theory values (k = 1, ..., Npt);
N sk is the stat.+syst. uncorrelated error; λα are sources of syst. errorsPavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 49
χ2 values per experiment (PRELIMINARY)
PDF set Order All Combined BCDMS CDF, D0 D0 Run-2 Aech,
expts. HERA-1 Fp,d2 Run-2 1-jet pe
T > 25 GeV
CT10.1 1.11 1.17 1.10 1.33 3.72MSTW08 NLO 1.42 1.73 1.16 1.31 11.38
(1.28) (1.4) (1.17)NNPDF2.0 1.37 1.32 1.28 1.57 2.79
CT10.2 1.13 1.12 1.14 1.23 2.59MSTW08 1.34 1.36 1.15 1.38 9.84
NNPDF2.1 NNLO 1.57 1.36 1.30 1.51 5.45ABM’09 (5f) 1.65 1.4 1.49 2.63 23.78
HERA1.5 1.71 1.15 1.87 ? 5.4
Npoints 2798 579 590 182 12
Pavel Nadolsky (SMU) Galileo Galilei Institute September 23, 2011 50