(NN)PDFs, the LHC & the LHeCAlberto GuffantiNiels Bohr International Academy & Discovery CenterNiels Bohr Intitute - Copenhagen
Parton Distribution FunctionsWhat are they?
Consider a process with a single hadron in the initial state
The cross section for such a process can be written (Factorization Theorem) as
dσ = dξξ0
1
∫a∑ Da (x,µ
2 )dσ axξ, sµ2 ,α s (µ
2 )⎛⎝⎜
⎞⎠⎟+O 1
Qp
⎛⎝⎜
⎞⎠⎟
Partonic Cross Section
Parton Distribution Function
Parton Distribution FunctionsWhat are they?
• Parton Distribution Functions are non-perturbative objects and their value at given x and Q2 cannot be computed in QCD Perturbation Theory (Lattice?)
• ... but the scale dependence of PDFs is governed by the DGLAP evolution equations
∂qi (x,µ2 )
∂lnµ2 = α s (µ2 )
2πPqq (x)⊗ qi (x,µ
2 )⎡⎣ ⎤⎦ +α s (µ
2 )2π
Pqg ⊗ g(x,µ2 )⎡⎣ ⎤⎦
∂g(x,µ2 )∂lnµ2 = α s (µ
2 )2π
Pgq (x)⊗ qi (x,µ2 )+ qi (x,µ
2 )( )i∑⎡
⎣⎢⎤⎦⎥+ α s (µ
2 )2π
Pgg ⊗ g(x,µ2 )⎡⎣ ⎤⎦
• ... where the splitting functions (Pij) can be computed in Perturbation Theory and are known up to NNLO
[LO - Dokshitzer; Gribov, Lipatov; Altarelli, Parisi (1977)][NLO - Floratos, Ross, Sachrajda; Gonzalez-Arroyo, Lopez, Yndurain; Curci, Furmanski, Petronzio (1981)]
[NNLO - Moch, Vermaseren, Vogt (2004)]
The PDF fitting gameThe players
Collaboration Authors arXiv
ABM
CTEQ/TEA
GJR
HERAPDF
MSTW
NNPDF
S. Alekhin, J. Blümlein, S. Moch1105.5349, 1101.5261, 1107.3657,
0908.3128, 0908.2766, …
M. Guzzi J. Huston, H.-L. Lai, P. Nadolsky, J. Pumplin, D. Stump, C.-P. Yuan
1108.5112, 1101.0561, 1007.2241, 1004.4624, 0910.4183, 0904.2424, 0802.0007, …
M. Glück, P. Jimenez-Delgado, E. Reya 1003.3168, 0909.1711, 0810.4274, …
H1 and ZEUS Collaborations 1107.4193, 1006.4471, 0906.1108, …
A. Martin, J. Stirling, R. Thorne, G. Watt 1107.2624, 1006.2753, 0905.3531, 0901.0002, …
R. D. Ball, V. Bertone, S. Carrazza, F. Cerutti, C. S. Deans, L. Del Debbio, S. Forte, AG,
N. P. Hartland, J. I. Latorre, J. Rojo, M. Ubiali
1110.2483, 1108.2758, 1107.2652, 1103.2369, 1102.3182, 1101.1300, 1005.0397, 1002.4407,
0912.2276, 0906.1958, …
The PDF fitting gameStatus of PDF fits
DATASET PERT. ORDER
HQ TREATMENT αs PARAM. UNCERT.
ABM11
CT10
JR09
HERAPDF1.5
MSTW08
NNPDF2.3
DISDrell-Yan NLO
NNLOFFN
(BMSN)Fit
(multiple values available)
6 indep. PDFsPolynomial(25 param.)
Hessian(Δ𝛘2=1)
GlobalLO
NLONNLO
GM-VFNS(S-ACOT)
External(multiple values
available)
6 indep. PDFsPolynomial(26 param.)
Hessian(Δ𝛘2=100)
DISDrell-Yan
JetsNLO
NNLOFFNVFN Fit
5 indep. PDFsPolynomial(15 param.)
Hessian(Δ𝛘2=1)
DIS (HERA) NLONNLO
GM-VFNS(TR)
External(multiple values
available)
5 indep. PDFsPolynomial(14 param.)
Hessian(Δ𝛘2=1)
GlobalLO
NLONNLO
GM-VFNS(TR)
Fit(multiple values
available)
7 indep. PDFsPolynomial(20 param.)
Hessian(Δ𝛘2≃25)
GlobalLO
NLONNLO
GM-VFNS(FONLL)
External(multiple values
available)
7 indep. PDFsNeural Nets(259 param.)
Monte Carlo
The PDF fitting gameStatus of PDF fits - parton luminosities
Good compatibility among global fitsDifferences are more marked when comparing to restricted dataset fits
Parton Luminosities at 8 TeV
15 Juan Rojo QCD@LHC Workshop, 20/08/2012
XM210 310
Glu
on -
Glu
on L
umin
osity
0.80.850.9
0.951
1.051.1
1.151.2
1.251.3
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
NNPDF2.3 NNLO
CT10 NNLO
MSTW2008 NNLO
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
XM210 310
Glu
on -
Glu
on L
umin
osity
0.80.850.9
0.951
1.051.1
1.151.2
1.251.3
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
NNPDF2.3 NNLO
ABM11 NNLO
HERAPDF1.5 NNLO
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
Gluon-Gluon
Reasonable agreement for between global PDF sets (but larger differences near M=125 GeV) Larger differences with non-global sets: ABM11 softer gg luminosity, specially at large
masses, HERAPDF1.5 similar central values but larger PDF uncertainties
The PDF fitting gameStatus of PDF fits - parton luminosities
Good compatibility among global fitsDifferences are more marked when comparing to restricted dataset fits
Parton Luminosities at 8 TeV
15 Juan Rojo QCD@LHC Workshop, 20/08/2012
XM210 310
Glu
on -
Glu
on L
umin
osity
0.80.850.9
0.951
1.051.1
1.151.2
1.251.3
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
NNPDF2.3 NNLO
CT10 NNLO
MSTW2008 NNLO
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
XM210 310
Glu
on -
Glu
on L
umin
osity
0.80.850.9
0.951
1.051.1
1.151.2
1.251.3
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
NNPDF2.3 NNLO
ABM11 NNLO
HERAPDF1.5 NNLO
= 0.119S_LHC 8 TeV - Ratio to NNPDF2.3 NNLO,
Gluon-Gluon
Reasonable agreement for between global PDF sets (but larger differences near M=125 GeV) Larger differences with non-global sets: ABM11 softer gg luminosity, specially at large
masses, HERAPDF1.5 similar central values but larger PDF uncertainties
The PDF fitting gameStatus of PDF fits - LHC cross-section
LHC cross sections at 8 TeV
18 Juan Rojo QCD@LHC Workshop, 20/08/2012
) [pb
]+
(Wm
6.6
6.8
7
7.2
7.4
7.6
7.8 = 0.119S_) - VRAP NNLO - +(WmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
) [pb
]0
(Zm
1.061.08
1.11.121.141.161.181.2
1.221.241.26
= 0.119S_) - VRAP NNLO - 0(ZmLHC 8 TeV NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
CMS
Good agreement between all PDFs but ABM11, which leads to larger cross sections
CMS data disfavor ABM11, but experimental errors still large
W+
Z0
More stringent constraints from 8 TeV differential distributions
18 Juan Rojo QCD@LHC Workshop, 20/08/2012
) [pb
]+
(Wm
6.6
6.8
7
7.2
7.4
7.6
7.8 = 0.119S_) - VRAP NNLO - +(WmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
) [pb
]0
(Zm
1.061.08
1.11.121.141.161.181.2
1.221.241.26
= 0.119S_) - VRAP NNLO - 0(ZmLHC 8 TeV NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
CMS
Good agreement between all PDFs but ABM11, which leads to larger cross sections
CMS data disfavor ABM11, but experimental errors still large
W+
Z0
More stringent constraints from 8 TeV differential distributions
LHC cross sections at 8 TeV
19 Juan Rojo QCD@LHC Workshop, 20/08/2012
(tt) [
pb]
m
200
210
220
230
240
250
260
270 = 0.119S_+NNLL - approx(tt) - Top++ v1.2 NNLOmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
(tt) [
pb]
m
180
200
220
240
260
280+NNLL - ABM11approx(tt) - Top++ v1.2 NNLOmLHC 8 TeV
= 0.114s_ = 0.117s_ = 0.118s_ = 0.119s_
CMS data
CMS
Good agreement between all PDFs (but ABM systematically lower, even for !S = 0.119 )
CMS data clearly disfavor small values of !S . Use these data to extract !S from the total cross section?
More stringent constraints on PDFs from differential distributions from ATLAS and CMS
Also cross section ratios between 8 TeV and 7 TeV very useful for PDF information (arXiv:1206.3557)
tT
tT
LHC data are already starting to discriminate among predictions
The PDF fitting gameStatus of PDF fits - LHC cross-section
LHC cross sections at 8 TeV
18 Juan Rojo QCD@LHC Workshop, 20/08/2012
) [pb
]+
(Wm
6.6
6.8
7
7.2
7.4
7.6
7.8 = 0.119S_) - VRAP NNLO - +(WmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
) [pb
]0
(Zm
1.061.08
1.11.121.141.161.181.2
1.221.241.26
= 0.119S_) - VRAP NNLO - 0(ZmLHC 8 TeV NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
CMS
Good agreement between all PDFs but ABM11, which leads to larger cross sections
CMS data disfavor ABM11, but experimental errors still large
W+
Z0
More stringent constraints from 8 TeV differential distributions
18 Juan Rojo QCD@LHC Workshop, 20/08/2012
) [pb
]+
(Wm
6.6
6.8
7
7.2
7.4
7.6
7.8 = 0.119S_) - VRAP NNLO - +(WmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
) [pb
]0
(Zm
1.061.08
1.11.121.141.161.181.2
1.221.241.26
= 0.119S_) - VRAP NNLO - 0(ZmLHC 8 TeV NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
CMS
Good agreement between all PDFs but ABM11, which leads to larger cross sections
CMS data disfavor ABM11, but experimental errors still large
W+
Z0
More stringent constraints from 8 TeV differential distributions
LHC cross sections at 8 TeV
19 Juan Rojo QCD@LHC Workshop, 20/08/2012
(tt) [
pb]
m
200
210
220
230
240
250
260
270 = 0.119S_+NNLL - approx(tt) - Top++ v1.2 NNLOmLHC 8 TeV
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5CMS 2012
CMS
(tt) [
pb]
m
180
200
220
240
260
280+NNLL - ABM11approx(tt) - Top++ v1.2 NNLOmLHC 8 TeV
= 0.114s_ = 0.117s_ = 0.118s_ = 0.119s_
CMS data
CMS
Good agreement between all PDFs (but ABM systematically lower, even for !S = 0.119 )
CMS data clearly disfavor small values of !S . Use these data to extract !S from the total cross section?
More stringent constraints on PDFs from differential distributions from ATLAS and CMS
Also cross section ratios between 8 TeV and 7 TeV very useful for PDF information (arXiv:1206.3557)
tT
tT
20 Juan Rojo QCD@LHC Workshop, 20/08/2012
(H) [
pb]
m
19.5
20
20.5
21
21.5
22
22.5 = 0.119S_LHC 8 TeV - iHixs 1.3 NNLO -
NNPDF2.3MSTW08CT10ABM11HERAPDF1.5
For a Higgs boson of 125 GeV, and a common !S = 0.119 value, all PDF sets within the PDF4LHC envelope
ABM11 in agreement with global PDF sets for same !S value
HERAPDF1.5 errors larger due to gluon parametrization uncertainties in a DIS only fit. To be improved with HERA jets.
Available PDF4LHC recommendation did clearly a good job: updated PDF sets if anything have confirmed its validity
These results will be useful to discuss future PDF4LHC recommendations
gg->H
LHC data are already starting to discriminate among predictions
NNPDF MethodologyMain ingredients
Monte Carlo determination of uncertaintiesNo need to rely on linear propagation of errorsPossibility to test the impact of non-gaussianly distributed uncertaintiesPossibility to test for non-gaussian behaviour of uncertainties of fitted PDFs
Parametrization of PDFs using Neural NetworksProvide an unbiased parametrization
Determine the best fit PDFs using Cross-ValidationEnsures proper fitting, avoiding overlearning
NNPDF Methodology... in a Nutshell
Generate Nrep Monte Carlo replicas of the experimental data, taking into account all experimental correlations
Fit a set of Parton Distribution Functions, parametrized at the initial scale using Neural Networks, to each replica
Expectation values for observables are then given by
.... and corresponding formulae are used to compute uncertainties, correlations, etc.
Reweighting (NN)PDFsAssessing the impact of new data on PDF fits
[R. D. Ball et al., arXiv:1012.0836]
[R. D. Ball et al., arXiv:1108.1758]
The Nrep replicas of a NNPDF fit give the probability density in thespace of PDFs
Expectation values for observables computed as
hF [fi(x , Q2)]i =1
Nrep
NrepX
k=1
F⇣
f (net)(k)i (x , Q2)
⌘
(... the same is true for errors, correlations, etc.)
We can assess the impact of including new data in the fit updatingthe probability density distribution without refitting.
A. Guffanti (NBIA & Discovery Center) PDFs@LHC 54 / 69
NNPDF MethodologyMonte Carlo replicas generation
Monte Carlo replicas are generated according to the distribution
where ri are (gaussianly distributed) random numbers
Validate Monte Carlo replicas against experimental data
O(1000) replicas needed to reproduce correlations in experimental data to percent accuracy
NNPDF MethodologyMonte Carlo replicas generation
Generate artificial data according to distribution
O(art) (k)i = (1 + r (k)
N �N)
"O(exp)
i +
NsysX
p=1
r (k)p �i,p + r (k)
i,s �
is
#
where ri are univariate (gaussianly distributed) random numbers
Validate Monte Carlo replicas against experimental data(statistical estimators, faithful representation of errors, convergence rateincreasing Nrep)
O(1000) replicas needed to reproduce correlations to percent accuracy
A. Guffanti (NBIA & Discovery Center) NNPDF4LHC 13 / 40
NNPDF MethodologyMonte Carlo replicas generation
Generate artificial data according to distribution
O(art) (k)i = (1 + r (k)
N �N)
"O(exp)
i +
NsysX
p=1
r (k)p �i,p + r (k)
i,s �
is
#
where ri are univariate (gaussianly distributed) random numbers
Validate Monte Carlo replicas against experimental data(statistical estimators, faithful representation of errors, convergence rateincreasing Nrep)
O(1000) replicas needed to reproduce correlations to percent accuracy
A. Guffanti (NBIA & Discovery Center) NNPDF4LHC 13 / 40
Reweighting (NN)PDFsThe reweighting idea
The N replicas of an NNPDF fit give the probability density in the space of PDFs
Expectation values for observables are computed as
We can then assess the impact of including new data in the fit updating the probability density without performing a complete refit
Reweighting (NN)PDFsAssessing the impact of new data on PDF fits
[R. D. Ball et al., arXiv:1012.0836]
[R. D. Ball et al., arXiv:1108.1758]
The Nrep replicas of a NNPDF fit give the probability density in thespace of PDFs
Expectation values for observables computed as
hF [fi(x , Q2)]i =1
Nrep
NrepX
k=1
F⇣
f (net)(k)i (x , Q2)
⌘
(... the same is true for errors, correlations, etc.)
We can assess the impact of including new data in the fit updatingthe probability density distribution without refitting.
A. Guffanti (NBIA & Discovery Center) PDFs@LHC 54 / 69
[R. D. Ball et al, arXiv:1012.0836][R. D. Ball et al, arXiv:1108.1758]
We can apply Bayes Theorem to determine the conditional probability of the PDF upon inclusion of the new data
Averages over the sample are no weighted sums
and the weights are given by
Reweighting (NN)PDFsThe reweighting formula
[R. D. Ball et al, arXiv:1012.0836][R. D. Ball et al, arXiv:1108.1758]
Reweighting (NN)PDFsAssessing the impact of new data on PDF fits
[R. D. Ball et al., arXiv:1012.0836]
[R. D. Ball et al., arXiv:1108.1758]
According to Bayes Theorem we have
Pnew({f}) = N�P(�2|{f})Pinit({f}) , P(�2|{f}) = [�2(y , {f})]ndat�1
2 e��2(y,{f})2
Averages over the sample are now weighted sums
hF [fi(x , Q2)]i =NrepX
k=1
wkF⇣
f (net)(k)i (x , Q2)
⌘
where the weights are
wk =[�2(y , fk )]
ndat�12 e��2(y,fk )
2
PNrepi=1 [�2(y , fi)]
ndat�12 e��2(y,fi )
2
A. Guffanti (NBIA & Discovery Center) PDFs@LHC 55 / 69
Reweighting (NN)PDFsAssessing the impact of new data on PDF fits
[R. D. Ball et al., arXiv:1012.0836]
[R. D. Ball et al., arXiv:1108.1758]
According to Bayes Theorem we have
Pnew({f}) = N�P(�2|{f})Pinit({f}) , P(�2|{f}) = [�2(y , {f})]ndat�1
2 e��2(y,{f})2
Averages over the sample are now weighted sums
hF [fi(x , Q2)]i =NrepX
k=1
wkF⇣
f (net)(k)i (x , Q2)
⌘
where the weights are
wk =[�2(y , fk )]
ndat�12 e��2(y,fk )
2
PNrepi=1 [�2(y , fi)]
ndat�12 e��2(y,fi )
2
A. Guffanti (NBIA & Discovery Center) PDFs@LHC 55 / 69
Reweighting (NN)PDFsAssessing the impact of new data on PDF fits
[R. D. Ball et al., arXiv:1012.0836]
[R. D. Ball et al., arXiv:1108.1758]
According to Bayes Theorem we have
Pnew({f}) = N�P(�2|{f})Pinit({f}) , P(�2|{f}) = [�2(y , {f})]ndat�1
2 e��2(y,{f})2
Averages over the sample are now weighted sums
hF [fi(x , Q2)]i =NrepX
k=1
wkF⇣
f (net)(k)i (x , Q2)
⌘
where the weights are
wk =[�2(y , fk )]
ndat�12 e��2(y,fk )
2
PNrepi=1 [�2(y , fi)]
ndat�12 e��2(y,fi )
2
A. Guffanti (NBIA & Discovery Center) PDFs@LHC 55 / 69
Reweighting (NN)PDFsValidating the reweighting procedure
We started from a fit to DIS and Drell-Yan data and included Tevatron inclusive jet data (CDF & D0) via refitting and reweighting
Reweighting with the two dataset at the same time or separately (in either order) yields identical results
Reweighting and refitting yield statistically equivalent results in the region constrained by the new data
CDF D0 CDF+D0Data points 76 110 186
Neff 290.8 565.8 334.5
Table 1: Datasets used in the Tevatron Run II inclusive jet reweighting exercise. For eachset the number of data points and the effective number of replicas of the reweighted setof Nrep = 1000 replicas are given.
x0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7
] 02xg
[x,Q
0
0.2
0.4
0.6
0.8
1Prior (NNPDF 2.0 DIS+DY)Refitted (NNPDF 2.0)W(CDF+D0)W(CDF)W(D0)W(D0)W(CDF)
)20
CDF/D0 Inclusive Jets - xg(x,Q
x0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7
] 02xg
[x,Q
-1
-0.5
0
0.5
1
1.5
2
2.5
3Prior (NNPDF 2.0 DIS+DY)Refitted (NNPDF 2.0)W(CDF+D0)W(CDF)W(D0)W(D0)W(CDF)
), ratio to NNPDF2.0 DIS+DY20
CDF/D0 Inclusive Jets - xg(x,Q
Figure 6: Comparison of the large-x gluon PDF for prior set, reweighted sets with differentsuccessive reweighting orders and refitted set, when the jet data of Table 1 are included inthe NNPDF2.0 NLO DIS+DY fit. Results are shown at Q2 = 2 GeV2, both in absolutescale (left) and as a ratio to the prior (right).
4.2 Tevatron Inclusive Jets
The first exercise we present is an extension of the reweighting proof-of-concept in Section 4of [14]. There, Run II Tevatron inclusive jet data production were included by reweightinga PDF set extracted from a NLO fit to DIS and Drell-Yan data (NNPDF2.0 DIS+DY) andthe results compared to those obtained from a fit which included the same DIS, Drell-Yanand inclusive jet datasets all treated in the same way (NNPDF2.0).
In this Section we look again at the inclusion via reweighting of the same datasets,namely the CDF Run II-kt and D0 Run II-cone inclusive jet data in the NNPDF2.0DIS+DY fit, but we now focus on comparing the results obtained in the following twocases:
(a) the two new datasets are included by reweighting the prior fit in a single step withboth datasets;
(b) one of the datasets is included by reweighting, an unweighted set of PDFs is con-structed using the procedure detailed in Section 3, and finally the latter set isreweighted again with the second dataset.
We will carry out the successive reweighting procedure (b) twice, exchanging the orderin which the CDF and D0 datasets are included, in order to test the commutativity ofthe procedure. A final unweighting is performed for all the reweighted sets and the PDFcomparisons and computations of distances are performed using these unweighted sets.
The number of data points and the effective number of replicas Neff after reweightingwith these data of a set of Nrep = 1000 replicas are summarized in Table 1. In each
16
NNPDF2.3 Dataset
x-510 -410 -310 -210 -110 1
]2 [
GeV
2 T /
p2
/ M
2Q
1
10
210
310
410
510
610
710NMC-pdNMCSLACBCDMS
HERAI-AVCHORUSFLH108
NTVDMNZEUS-H2ZEUSF2CH1F2CDYE605DYE886CDFWASYCDFZRAPD0ZRAP
CDFR2KTD0R2CONATLAS-WZ-36pbCMS-WEASY-840PBLHCB-WZ-36pbATLAS-JETS-10
NNPDF2.3 dataset Experiment DataFixed Target DIS 1952
HERA DIS 834Fixed Target DY 318
Tevatron W/Z 70Tevatron Jets 186
LHC W/Z 56LHC Jets 90
3506 data points(in the NNLO global fit)
NNPDF2.3The LHC data - Fit quality
NLO NNPDF2.3 noLHC NNPDF2.3NMCpd
NMCSLAC
BCDMSHERA-I
CHORUSNuTeV
DYE605DYE866
CDFWASYCDFZRAPD0ZRAP
ATLAS-WZCMS-WEASY
LHCb-WZCDFR2KTD0R2CON
ATLAS-JETS-2010
0.93 0.931.59 1.611.28 1.261.20 1.191.01 1.001.09 1.100.42 0.450.85 0.881.24 1.281.45 1.541.77 1.790.57 0.571.37 1.271.50 1.041.24 1.210.60 0.610.84 0.841.57 1.55
Compare the quality of the fit to LHC data before and after inclusion in the global fit
Including LHC data in the fit improves the quality of their description, w/o deteriorating quality of the fit to other datasets
Moderate impact of the LHC data, supporting consistency of the global fit framework
Fit quality is comparable at NLO and NNLO, thought the former marginally better
NNPDF2.3The LHC data - Fit quality
NNLO NNPDF2.3 noLHC NNPDF2.3NMCpd
NMCSLAC
BCDMSHERA-I
CHORUSNuTeV
DYE605DYE866
CDFWASYCDFZRAPD0ZRAP
ATLAS-WZCMS-WEASY
LHCb-WZCDFR2KTD0R2CON
ATLAS-JETS-2010
0.94 0.941.56 1.571.04 1.021.28 1.291.03 1.011.07 1.060.48 0.551.07 1.021.61 1.621.66 1.702.15 2.120.64 0.631.94 1.461.37 0.961.33 1.220.67 0.670.94 0.931.45 1.42
Compare the quality of the fit to LHC data before and after inclusion in the global fit
Including LHC data in the fit improves the quality of their description, w/o deteriorating quality of the fit to other datasets
Moderate impact of the LHC data, supporting consistency of the global fit framework
Fit quality is comparable at NLO and NNLO, thought the former marginally better
NNPDF2.3Collider fit - are we there yet?
It is the fit we would love to haveOnly high energy data: minimize the effects of higher-twist contributionsOnly proton data: no assumptions based on models for nuclear corrections
x0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.1
0.2
0.3
0.4
0.5
0.6
0.7)
02 (x, Q3xT
NNPDF2.3 NLO
NNPDF2.3 NLO Collider-Only
) 02 (x, Q3xT
x-410 -310 -210 -110
0
1
2
3
4
5 )
02xg ( x, Q
NNPDF2.3 NLO
NNPDF2.3 NLO Collider-Only
)02xg ( x, Q
Gluon distribution is very well constrained both at small-x (HERA) and large-x (Tevatron/LHC jets)
PDF combinations sensitive to light flavour separationhave substantially larger uncertainties (missing constraints from fixed target DIS/DY data)
Uncertainties on “fixed target” observables are still unacceptably large
... things can only get better with more LHC data coming (W+c, low mass DY, photons, high pt Z/W ...)
NNPDF2.3Phenomenology - parton luminosities
Reduction in uncertainty on gluon-gluon luminosity for larger final state invariant masses when going from NNPDF2.1 to NNPDF2.3
NNPDF2.3 quark-antiquark luminosity at large invariant masses somewhat smaller than NNPDF2.1
NNPDF2.3Phenomenology - W/Z production
Mostly sensitive to quark parton luminositiesPredictions from NNPDF2.3 sets are compatible with each other and with predictions obtained using the NNPDF2.1 global setLargest differences with collider only fit, although the latter has larger uncertainties
[nb]
il)B+
(Wm
5.5
6
6.5
7
7.5NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
il)B-
(Wm
3.8
4
4.2
4.4
4.6
4.8
5
5.2NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
ll)B0
(Zm
0.8
0.9
1
1.1
1.2
NNPDF + VRAPNLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
(H) [
pb]
m
8
10
12
14
16
18
20
NNPDF + iHixs 1.3NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
(tt) [
pb]
m
140
160
180
200
220
240
NNPDF + Top++NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
[nb]
il)B+
(Wm
5.5
6
6.5
7
7.5NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
il)B-
(Wm
3.8
4
4.2
4.4
4.6
4.8
5
5.2NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
ll)B0
(Zm
0.8
0.9
1
1.1
1.2
NNPDF + VRAPNLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
(H) [
pb]
m
8
10
12
14
16
18
20
NNPDF + iHixs 1.3NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
(tt) [
pb]
m
140
160
180
200
220
240
NNPDF + Top++NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
NNPDF2.3Phenomenology - top/Higgs production
[nb]
il)B+
(Wm
5.5
6
6.5
7
7.5NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
il)B-
(Wm
3.8
4
4.2
4.4
4.6
4.8
5
5.2NNPDF + VRAP
NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll
[nb]
ll)B0
(Zm
0.8
0.9
1
1.1
1.2
NNPDF + VRAPNLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.32.3noLHC
2.3coll
2.1 2.32.3noLHC
2.3coll(H
) [pb
]m
8
10
12
14
16
18
20
NNPDF + iHixs 1.3NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
(tt) [
pb]
m
140
160
180
200
220
240
NNPDF + Top++NLO, 7 TeVNLO, 8 TeVNNLO, 7 TeVNNLO, 8 TeV
2.1 2.3 2.3noLHC2.3coll
2.1 2.3 2.3noLHC2.3coll
Mostly sensitive to quark parton luminositiesPredictions from NNPDF2.3 sets are compatible with each other and with predictions obtained using the NNPDF2.1 global setLargest differences with collider only fit, although the latter has larger uncertainties
The LHCLepton-hadron collision at the LHC
The LHeC is a proposed facility at CERN, where a (7 TeV) LHC proton beam is brought into collision with a lepton beam
The LHeC is designed to operate simultaneously with the LHC
Two options for the lepton beam are considered
Linac-Ring optionRing-Ring option
Many more details can be found in the recently published Conceptual Design Report (arXiv:1206.2913)
ICHEP&LHeC&Max&Klein&7.7.2012& 1&
The&LHeC&Project&&&
Max&Klein&University&of&Liverpool&
&For&the&LHeC&Study&Group&
&Brief&overview&on&the&&Status&and&Prospects&
&&&&
at&this&conference:&A.Polini&–&Detector&C.Glasman&–&QCD&P.Newman&–&eA&U.Klein&–&Higgs&
&ICHEP&2012&Melbourne&7.7.&
The LHCPotential to constrain PDFs
The measurements at the LHeC will nicely complement the pp and pA measurements from the LHC experiments
Unique possibility to explore the small-x region in DIS
Precise measurements of the Neutral and Charrged Current DIS cross-section at large-x for accurate flavour separation
Precise determination of heavy flavour parton distributions
ComplemenNng$the$LHC$with$ep/A$LHC$partons:$W,Z$+c,b$new$constraints$but$severely$limited$in$x,Q2$range$
Discoveries$at$the$LHC$will$be$at$high$masses:$large$x$and$very$high$Q2$which$require$high$s,$lumi$of$LHeC$for$precision$PDFs$(u,d,xg$mainly)$
If$the$Higgs$exists,$its$study$will$become$a$major$field$of$research:$ep:$$WW$H$$bbar$(CP$odd/even?)$
top$distribuNon$in$the$proton$TDF$
IF$RP$is$violated$and$LQ$or$RPV$SUSY$discovered:$LHeC$is$uniquely$suited$
AA:$QGP:$study$iniNal$state$in$eA$Resolve$parton$distribuNons$in$nuclei$
LHeC$is$unique$in$various$areas,$e.g.:$$Low$x$and$saturaNon$physics$Strong$coupling$constant$to$0.1%$level$
In$Drell]Yan$kinemaNcs:$mass$and$rapidity$relate$to$Q2$and$x$
DIS$
$
The LHCPotential to constrain PDFs
The measurements at the LHeC will nicely complement the pp and pA measurements from the LHC experiments
Unique possibility to explore the small-x region in DIS
Precise measurements of the Neutral and Charrged Current DIS cross-section at large-x for accurate flavour separation
Precise determination of heavy flavour parton distributions
Partons, QCD and Low x Physics at the LHeC 10
LHeC physics: proton PDFs• PDFs are extracted from fits to data assuming a• functional form for x at a given Q2
min value and then• evolved via DGLAP evolution equations• Current PDF status from HERA:
0
0.2
0.4
0.6
0.8
1
-410 -310 -210 -110 10
0.2
0.4
0.6
0.8
1
HERAPDF1.5 NNLO (prel.)
exp. uncert.
model uncert. parametrization uncert.
x
xf 2 = 2 GeV2Q
vxu
vxd
0.05)×xS ( 0.05)×xg (
HER
APD
F St
ruct
ure F
unct
ion
Wor
king
Gro
upM
arch
201
1
H1 and ZEUS HERA I+II PDF Fit
0
0.2
0.4
0.6
0.8
1 • Impact of LHeC• data expected to• be large thanks to− new kinematic− range− huge luminosity− polarised beams− deuteron beams− high precision data
⇒ LHeC has the potential to provide significant⇒ LHeC has the potential to provide significant⇒ constraints to the PDFs →
u valence
d valence
gluon
Q2=1.9 GeV2
high x
• The reduction of uncertainties in nuclear PDFs are• presented in the talk on heavy ions
LHeC constraint
Claudia Glasman (Universidad Autonoma de Madrid) ICHEP2012 (Melbourne), 4-11 July 2012
Partons, QCD and Low x Physics at the LHeC 10
LHeC physics: proton PDFs• PDFs are extracted from fits to data assuming a• functional form for x at a given Q2
min value and then• evolved via DGLAP evolution equations• Current PDF status from HERA:
0
0.2
0.4
0.6
0.8
1
-410 -310 -210 -110 10
0.2
0.4
0.6
0.8
1
HERAPDF1.5 NNLO (prel.)
exp. uncert.
model uncert. parametrization uncert.
x
xf 2 = 2 GeV2Q
vxu
vxd
0.05)×xS ( 0.05)×xg (
HER
APD
F St
ruct
ure
Func
tion
Wor
king
Gro
upM
arch
201
1
H1 and ZEUS HERA I+II PDF Fit
0
0.2
0.4
0.6
0.8
1 • Impact of LHeC• data expected to• be large thanks to− new kinematic− range− huge luminosity− polarised beams− deuteron beams− high precision data
⇒ LHeC has the potential to provide significant⇒ LHeC has the potential to provide significant⇒ constraints to the PDFs →⇒ A precise value of αs from DIS will also be possible⇒ at LHeC
u valence
d valence
gluon
Q2=1.9 GeV2
low x
LHeC constraint
Claudia Glasman (Universidad Autonoma de Madrid) ICHEP2012 (Melbourne), 4-11 July 2012
Conclusions & Outlook
Parton Distribution Function are an essential ingredient of theoretical predictions for observables at hadron-hadron and lepton-hadron colliders
The NNPDF methodology is ideally suited to tackle the shortcomings of the standard PDF fitting methodology
The reweighting technique can be used to quickly and reliably assess the impact of new data in parton distribution functions fits without the need for refitting
NNPDF2.3 is the first global PDF fit to include LHC data
The proposed LHeC experiment will be a unique tool to understand the structure on the proton and accurately determine parton densities
Top Related