Probing the large-x gluon PDF at NNLO with top-pair di...
Transcript of Probing the large-x gluon PDF at NNLO with top-pair di...
Probing the large-x gluon PDF at NNLOwith top-pair differential data
based on JHEP 1704 (2017) 044with M. Czakon, A. Mitov, N.P. Hartland and J.Rojo
Heavy Flavour Production at the LHC
Emanuele R. Nocera
Rudolf Peierls Centre for Theoretical Physics - University of Oxford
IPPP Durham – 6th September 2017
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 1 / 21
Unveiling a precision large-x gluon PDFThe gluon PDF at medium to large x plays a leading role in many BSM scenarios
(gluino pair production, Kaluza-Klein graviton production, resonances in the mtt̄ spectrum, . . . )
Usual constraints come from inclusive top and jet cross sections (Tevatron and LHC)
inclusive top
complete NNLO QCD corrections[PRL 109 (2012) 132001] [JHEP 1301 (2013) 1380]
[PRL 110 (2013) 252004]
inclusive jets
x0 0.1 0.2 0.3 0.4 0.5 0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
Ratio to NNPDF3.0 NNLO no jetsRatio to NNPDF3.0 NNLO no jets
NNPDF3.0 no jets
NNPDF3.02 = 100 GeV2Q
Ge
ne
ra
ted
wit
h A
PF
EL
2.7
.1 W
eb
complete NNLO QCD corrections[PRL 110 (2013) 162003] [JHEP 1401 (2014) 110]
[PRL 118 (2017) 072002]
Dedicated phenomenological studies [JHEP 1307 (2013) 167] [JHEP 1410 (2014) 145]
Data on inclusive top and jet production are routinely included in global PDF fits
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 2 / 21
A new ingredient: tt̄ differential distributions (` + jets)Complete NNLO QCD corrections for stable top quarks, with scale optimization[PRL 115 (2015) 052001; PRL 115 (2016) 082003; JHEP 05 (2016) 034; JHEP 1704 (2017) 071; PRD 94 (2016) 114033]
Precise data from ATLAS and CMS at√s = 8 TeV, with a full breakdown of systematics
[EPJ C76 (2016) 538; EPJ C75 (2015) 542]
Dataset Ndat Kinematics
ATLAS dσ/dptT 8 0 < ptT < 500 GeV
ATLAS dσ/d|yt| 5 0 < |yt| < 2.5
ATLAS dσ/d|ytt̄| 5 0 < |ytt̄| < 2.5
ATLAS dσ/dmtt̄ 7 345 < mtt̄ < 1600 GeV
ATLAS (1/σ)dσ/dptT 8 0 < ptT < 500 GeV
ATLAS (1/σ)dσ/d|yt| 5 0 < |yt| < 2.5
ATLAS (1/σ)dσ/d|ytt̄| 5 0 < |ytt̄| < 2.5
ATLAS (1/σ)dσ/dmtt̄ 7 345 < mtt̄ < 1600 GeV
Dataset Ndat Kinematics
CMS dσ/dptT 8 0 < ptT < 500 GeV
CMS dσ/dyt 10 −2.5 < yt < 2.5
CMS dσ/dytt̄ 10 −2.5 < ytt̄ < 2.5
CMS dσ/dmtt̄ 7 345 < mtt̄ < 1600 GeV
CMS (1/σ)dσ/dptT 8 0 < ptT < 500 GeV
CMS (1/σ)dσ/dyt 10 −2.5 < yt < 2.5
CMS (1/σ)dσ/dytt̄ 10 −2.5 < ytt̄ < 2.5
CMS (1/σ)dσ/dmtt̄ 7 345 < mtt̄ < 1600 GeV
Normalised and absolute data is provided for ATLASCMS provides normalised data; absolute data reconstructed from inclusive cross section(If normalised distributions are included in a fit, the corresponding total cross section is included)
Correlations among different distributions not provided(Only one distribution per experiment can be included in a fit, avoid multiple counting)
Are the various datasets compatible? What is the level of PDF constraint?Is normalising beneficial? Which combination of distributions is the best?
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 3 / 21
PDF sensitivity to tt̄ differential distributions
Plot correlation coefficients of PDFs with absolute tt̄ differential distributions
ρ[A,B] =〈AB〉rep − 〈A〉rep〈B〉rep
σAσB
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[g(x,Q),dσ/dptT]
Q=100 GeV
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[g(x,Q),dσ/dyt]
Q=100 GeV
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[g(x,Q),dσ/dmtt-]
Q=100 GeV
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[g(x,Q),dσ/dytt-]
Q=100 GeV
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 4 / 21
PDF sensitivity to tt̄ differential distributions
Plot correlation coefficients of PDFs with absolute tt̄ differential distributions
ρ[A,B] =〈AB〉rep − 〈A〉rep〈B〉rep
σAσB
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[q(x,Q),dσ/dptT]
Q=100 GeV
u-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
u--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
c-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
b -1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[q(x,Q),dσ/dyt]
Q=100 GeV
u-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
u--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
c-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
b
-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[q(x,Q),dσ/dmtt-]
Q=100 GeV
u-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
u--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
c-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
b -1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
ρ[q(x,Q),dσ/dytt-]
Q=100 GeV
u-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
u--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
d--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
s--1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
c-1
-0.5
0
0.5
1
0.001 0.01 0.1 1
x
b
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 4 / 21
Theory configuration: NLO/NNLO computationAt NLO, use interpolated partonic cross section tables (APPLgrid/FK tables)
Matrix elements: (Sherpa × Openloops) via MCgrids
Stong coupling: set as per PDF set, e.g. αs(MZ) = 0.118 (NNPDF), αs(MZ) = 0.113 (ABM)
Top mass: set as per PDG average, mt = 173.3 GeV (mt sensitivity in [PRD 94 (2016) 114033])variation below 6h for yt and ytt̄ distributions, up to 4% close to threshold for ptT and mtt̄
upon ∆mt = 1 GeV
Choice of dynamical scales: set as per [arXiv:1606.03350]
yt, ytt̄, mtt̄ : µR = µF =1
4
(√m2t + ptT
2 +
√m2t + pt̄T
2)
ptT : µR = µF =1
2
√m2t + ptT
2 as average over t/t̄
At NNLO, a full computation remains prohibitively expensive for inclusion in a PDF fit
Use of bin-by-bin K-factors (C-factors) computed with identical settings as for NLO
C =σ̂NNLO ⊗ LNNLO
σ̂NLO ⊗ LNNLO
Total cross sections computed using top++ [CPC 185 (2014) 2930] at NNLO
integration of differential distributions with dynamical scales in close agreement withNNLO+NNLL top++ (∼ 2% higher than NNLO top++), but inconsequential for this study
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 5 / 21
Theory configuration: NNLO C-factors
0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
0
60
100
150
200
260
320
400
500
C-factor (dσ/dpTt)
NNPDF3.0CT14
MMHT14 0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
-2.5
-1.6
-1.2
-0.8
-0.4 0
0.4
0.8
1.2
1.6
2.5
C-factor (dσ/dyt)
NNPDF3.0CT14
MMHT14
0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
345
400
470
550
650
800
1100
1600
C-factor (dσ/dmtt)
NNPDF3.0CT14
MMHT14 0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
-2.5
-1.3
-0.9
-0.6
-0.3 0
0.3
0.6
0.9
1.3
2.5
C-factor (dσ/dytt)
NNPDF3.0CT14
MMHT14
C-factors are stable undervariation of applied PDF
ptT distribution:from 9% (low ptT ) to closeto unity (ptT ' 500 GeV)
mtt̄ distribution:from 5% (low mtt̄) to12% (mtt̄ & 1 TeV)
yt and ytt̄ distributions:around 6%-9%, almostflat in the data region
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 6 / 21
Data/Theory comparison: ptT
0
0.3
0.6
0.9
1.2
1.5
1.8
2.1 dσ/dptT [pb/GeV]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
0
60
100
150
200
260
320
400
500
ptT [GeV]
Ratio to NNPDF3.0
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008 (1/σ)dσ/dptT [1/GeV]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
0
60
100
150
200
260
320
400
500
ptT [GeV]
Ratio to NNPDF3.0
PDF set ATLAS dσ/dptT CMS dσ/dptT ATLAS (1/σ)dσ/dptT CMS (1/σ)dσ/dptT
NNPDF3.0 0.84 (0.66) 1.24 (0.91) 3.13 (0.94) 2.03 (0.51)MMHT14 0.63 (0.44) 1.54 (1.47) 2.23 (0.54) 3.15 (0.77)CT14 0.76 (0.42) 1.67 (1.77) 2.33 (0.62) 2.88 (0.70)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 7 / 21
Data/Theory comparison: mtt̄
0
0.2
0.4
0.6
0.8
1
1.2
1.4 dσ/dmtt- [pb/GeV]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
345
400
470
550
650
800
1100
1600
mtt- [GeV]
Ratio to NNPDF3.0
0
0.001
0.002
0.003
0.004
0.005
0.006 (1/σ)dσ/dmtt- [1/GeV]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
345
400
470
550
650
800
1100
1600
mtt- [GeV]
Ratio to NNPDF3.0
PDF set ATLAS dσ/dmtt̄ CMS dσ/dmtt̄ ATLAS (1/σ)dσ/dmtt̄ CMS (1/σ)dσ/dmtt̄
NNPDF3.0 0.77 (0.38) 5.73 (4.36) 1.57 (0.10) 10.6 (3.87)MMHT14 0.58 (0.24) 7.32 (5.74) 1.01 (0.05) 13.5 (4.93)CT14 0.61 (0.19) 7.28 (6.06) 1.09 (0.05) 13.5 (4.82)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 8 / 21
Data/Theory comparison: yt
20
40
60
80
100
120
140 dσ/dyt [pb]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
-2.5
-1.6
-1.2
-0.8
-0.4 0
0.4
0.8
1.2
1.6
2.5
yt
Ratio to NNPDF3.0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (1/σ)dσ/dytNNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
-2.5
-1.6
-1.2
-0.8
-0.4 0
0.4
0.8
1.2
1.6
2.5
yt
Ratio to NNPDF3.0
PDF set ATLAS dσ/dyt CMS dσ/dyt ATLAS (1/σ)dσ/dyt CMS (1/σ)dσ/dyt
NNPDF3.0 0.73 (0.28) 3.04 (1.05) 4.06 (2.85) 3.29 (1.49)MMHT14 1.36 (0.29) 2.12 (0.98) 12.1 (6.82) 2.40 (1.09)CT14 1.28 (0.20) 2.23 (1.47) 10.3 (5.71) 2.33 (0.96)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 9 / 21
Data/Theory comparison: ytt̄
20
40
60
80
100
120
140 dσ/dytt- [pb]
NNLO theory
NNPDF3.0MMHT14
CT14CMS
ATLAS
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
-2.5
-1.3
-0.9
-0.6
-0.3 0
0.3
0.6
0.9
1.3
2.5
ytt-
Ratio to NNPDF3.0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55 (1/σ)dσ/dytt-
NNLO theoryNNPDF3.0
MMHT14CT14CMS
ATLAS
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
-2.5
-1.3
-0.9
-0.6
-0.3 0
0.3
0.6
0.9
1.3
2.5
ytt-
Ratio to NNDPF3.0
PDF set ATLAS dσ/dytt̄ CMS dσ/dytt̄ ATLAS (1/σ)dσ/dytt̄ CMS (1/σ)dσ/dytt̄
NNPDF3.0 0.84 (0.21) 0.99 (0.74) 3.59 (1.48) 1.17 (0.75)MMHT14 2.36 (0.29) 2.27 (1.52) 15.6 (5.49) 3.33 (2.10)CT14 2.69 (0.19) 1.88 (1.67) 12.7 (5.26) 2.53 (1.51)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 10 / 21
Including tt̄ distributions in a fit: fit settingsQualitative tension between CMS and ATLAS data, use fits to investigate quantitatively
Fits are performed in the NNPDF3 framework, with perturbative charm
Dataset Fit ID1 2 3 4 5 6 7 8 9 10
Baseline + 2� 2� 2� 2� 2� 2� 2� 2� 2� 2�
ATLAS dσ/dptT 2�ATLAS dσ/dyt 2�ATLAS dσ/dytt̄ 2�ATLAS dσ/dmtt̄ 2�ATLAS (1/σ)dσ/dptT 2�ATLAS (1/σ)dσ/dyt 2�ATLAS (1/σ)dσ/dytt̄ 2�ATLAS (1/σ)dσ/dmtt̄ 2�ATLAS σtt̄ 2� 2� 2� 2� 2�
CMS dσ/dptT 2�CMS dσ/dyt 2�CMS dσ/dytt̄ 2�CMS dσ/dmtt̄ 2�CMS (1/σ)dσ/dptT 2�CMS (1/σ)dσ/dyt 2�CMS (1/σ)dσ/dytt̄ 2�CMS (1/σ)dσ/dmtt̄ 2�CMS σtt̄ 2� 2� 2� 2� 2�
Baseline 1: HERA-only (legacy combinations of inclusive and charm measurements)
Baseline 2: Global (same dataset as NNPDF3.0 without jet data)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 11 / 21
NNLO HERA-only fit quality: χ2/NdatDataset Fit ID one distribution
1 2 3 4 5 6 7 8 9 10 at a time
ATLAS dσ/dptT 2.30 2.48 0.73 3.16 3.46 2.04 1.34 3.28 4.88 2.89 0.44ATLAS dσ/dyt 0.82 1.14 1.21 1.06 0.75 1.04 1.31 0.59 0.75 0.74 0.47ATLAS dσ/dytt̄ 1.12 1.90 2.40 2.83 0.45 4.43 1.96 1.88 0.40 1.49 0.43ATLAS dσ/dmtt̄ 4.27 2.93 2.41 2.81 4.33 1.53 2.70 2.88 4.37 5.09 0.39ATLAS (1/σ)dσ/dptT 3.47 2.60 3.80 2.92 3.15 3.91 1.46 3.31 3.98 4.01 0.60ATLAS (1/σ)dσ/dyt 1.21 6.07 3.32 5.95 1.34 2.24 4.27 1.48 1.58 1.61 0.75ATLAS (1/σ)dσ/dytt̄ 3.11 12.8 5.09 8.34 0.72 7.04 4.95 3.60 0.53 2.60 0.45ATLAS (1/σ)dσ/dmtt̄ 8.14 3.07 6.53 4.94 5.42 20.5 6.44 5.61 4.40 3.03 0.55ATLAS σtt̄ 3.88 0.35 3.38 0.63 1.58 1.29 0.87 0.37 0.42 0.66
CMS dσ/dptT 2.04 2.29 0.82 3.29 2.99 1.52 1.44 2.81 4.16 2.32 0.82CMS dσ/dyt 3.38 2.48 2.91 1.75 3.51 3.47 2.32 3.03 3.48 4.81 1.30CMS dσ/dytt̄ 1.00 1.58 2.29 1.68 1.08 3.05 1.51 1.34 1.07 1.85 0.74CMS dσ/dmtt̄ 3.96 5.85 4.81 4.70 4.23 1.73 4.46 4.23 4.71 3.74 1.28CMS (1/σ)dσ/dptT 2.78 4.86 1.78 5.23 4.05 2.84 1.57 4.69 5.29 3.40 0.85CMS (1/σ)dσ/dyt 5.73 3.15 4.10 2.35 5.04 4.88 3.13 1.94 4.60 6.71 1.70CMS (1/σ)dσ/dytt̄ 1.68 2.27 2.62 2.11 1.40 3.42 1.78 1.49 1.20 1.98 0.75CMS (1/σ)dσ/dmtt̄ 5.30 10.3 7.83 8.24 7.06 2.71 7.45 7.41 8.06 6.26 0.92CMS σtt̄ 6.95 1.04 6.17 1.59 3.24 2.75 1.02 1.09 1.17 1.64
TOTAL 1.18 1.18 1.17 1.19 1.18 1.19 1.18 1.20 1.18 1.22
Overall satisfactory description of most of the fitted differential distributionsworst fit in the case of top rapidity and top-pair mass distributions
Disagreement from a tension between ATLAS and CMS data rather than from theoryinconsistency resolved by examining fits in which ATLAS and CMS data are included separately
The overall quality of the fit does not deteriorate upon the inclusion of tt̄ data
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 12 / 21
NNLO global fit quality: χ2/NdatDataset Fit ID one distribution no fixed-target
1 2 3 4 5 6 7 8 9 10 at a time DIS
ATLAS dσ/dptT 2.37 2.30 1.99 2.36 2.24 2.23 2.09 2.18 2.34 2.24ATLAS dσ/dyt 0.93 0.80 0.74 1.09 0.76 0.76 0.86 0.69 0.76 0.66ATLAS dσ/dytt̄ 2.44 2.03 1.96 2.59 1.32 2.32 2.11 1.74 1.26 1.80ATLAS dσ/dmtt̄ 4.27 4.47 4.68 4.14 4.92 4.02 4.34 4.79 4.98 4.99
ATLAS (1/σ)dσ/dptT 2.93 3.97 3.29 4.36 5.22 4.35 2.96 4.26 4.92 5.68 2.38 0.79ATLAS (1/σ)dσ/dyt 5.00 3.17 2.47 6.36 1.55 2.93 3.94 1.68 1.45 1.10 1.11ATLAS (1/σ)dσ/dytt̄ 9.69 5.59 5.89 8.95 2.68 5.73 6.73 3.57 2.17 3.73 1.12ATLAS (1/σ)dσ/dmtt̄ 2.30 2.80 3.31 2.67 3.96 4.21 3.09 3.68 3.77 2.98 1.88 0.61ATLAS σtt̄ 0.12 0.10 0.21 0.10 0.10 0.12 0.36 0.29 0.26 0.10
CMS dσ/dptT 3.50 3.46 2.60 3.50 3.03 3.00 2.85 3.11 3.24 2.92CMS dσ/dyt 3.48 3.71 4.05 2.66 4.18 3.49 3.38 4.23 4.43 4.99CMS dσ/dytt̄ 1.36 1.13 1.00 1.32 0.89 0.86 1.00 1.01 1.04 1.24CMS dσ/dmtt̄ 7.07 6.27 5.79 6.33 5.09 5.11 6.00 5.37 5.21 4.31
CMS (1/σ)dσ/dptT 4.31 4.00 3.39 4.28 3.65 3.59 3.56 3.57 3.73 3.48 3.03 0.90CMS (1/σ)dσ/dyt 3.66 4.10 4.45 3.10 4.98 4.06 3.65 4.76 5.13 6.09 1.66CMS (1/σ)dσ/dytt̄ 1.59 1.20 1.06 1.73 0.94 1.01 1.20 0.99 1.05 1.32 0.93CMS (1/σ)dσ/dmtt̄ 12.0 10.8 9.81 11.1 8.72 8.72 10.3 9.15 8.97 7.27 4.12 1.01CMS σtt̄ 0.10 0.05 0.26 0.19 0.32 0.21 0.11 0.10 0.15 0.35
TOTAL 1.20 1.19 1.20 1.20 1.19 1.21 1.20 1.21 1.20 1.21
Overall data description significantly worse than in HERA-only fits
Clear tensions between ATLAS and CMS data and the rest of the datasetinconsistency resolved in fits to individual ATLAS/CMS distributions, no fixed-target DIS data
The overall quality of the fit does not deteriorate upon the inclusion of tt̄ datatt̄ data has not enough weight to result in a deterioration in fit quality to the data in tension
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 13 / 21
Impact of tt̄ distributions on the gluon PDF at large x
0.001
0.01
0.1
1
10
100
0.001 0.01 0.1 1x
g(x,Q)
Q=100 GeV
HERA-onlyFIT 1FIT 2FIT 3FIT 4FIT 5FIT 6FIT 7FIT 8FIT 9
FIT 10
GlobalFIT 1FIT 2FIT 3FIT 4FIT 5FIT 6FIT 7FIT 8FIT 9
FIT 10FIT opt
0.001
0.01
0.1
0.3 0.4 0.5 0.6 0.7
x
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.001 0.01 0.1 1x
σg(x,Q)
Q=100 GeV
HERA-onlyFIT 1FIT 2FIT 3FIT 4FIT 5FIT 6FIT 7FIT 8FIT 9
FIT 10
GlobalFIT 1FIT 2FIT 3FIT 4FIT 5FIT 6FIT 7FIT 8FIT 9
FIT 10FIT opt
Fair degree of consistency in the impact of various distributions on the gluon PDFHERA-only fit (red): tt̄ data prefers a harder gluon w.r.t. to the baseline (solid red)
Global fit (green): tt̄ data prefers a softer gluon w.r.t. to the baseline (solid green)
Nice convergence/consistency check
Largest constraining power on the gluon PDF uncertainty in the global fitThere exists an optimal combination of tt̄ data that maximises this effect (bold dashed green)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 14 / 21
Impact of tt̄ distributions on the gluon PDF at large x
0.7
0.8
0.9
1
1.1
1.2
1.3 g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+dσ/dptT 0.7
0.8
0.9
1
1.1
1.2
1.3g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+σtt-+(1/σ)dσ/dp
tT
0.7
0.8
0.9
1
1.1
1.2
1.3 g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+dσ/dmtt- 0.7
0.8
0.9
1
1.1
1.2
1.3g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+σtt-+(1/σ)dσ/dmtt
-
0.7
0.8
0.9
1
1.1
1.2
1.3 g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+dσ/dyt 0.7
0.8
0.9
1
1.1
1.2
1.3g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+σtt-+(1/σ)dσ/dyt
0.7
0.8
0.9
1
1.1
1.2
1.3
0.01 0.1
x
g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+dσ/dytt-
0.01 0.1
0.7
0.8
0.9
1
1.1
1.2
1.3
x
g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global+σtt-
Global+σtt-+(1/σ)dσ/dytt
-
Significant reduction inthe gluon uncertainty
at large x
Affected kinematic regionas expected from the
correlation coefficients(0.1 . x . 0.7)
Gluon remarkablyconsistent in the fit acrossthe choice of distributions
Normalised distributionsappear to lead to a greaterreduction of uncertainties
Almost negligible impactof total inclusive cross
sections (in green)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 15 / 21
Procedural suggestionobservational facts
Impact of data on PDF central values appears relatively insensitive of distribution choiceall distributions are equal in the pool on the gluon PDF central value
While some distributions exhibit tensions, some appear perfectly compatibleselect distributions whose inclusion in the global fit leads to χ2/Ndat ∼ 1
Normalised distributions + total cross sections exhibit the largest constraining powerthey lead to the largest reduction of the gluon PDF uncertainties
Different distributions have their largest impact on a slightly different region of xchoose different distributions from ATLAS and CMS in order to maximise the kinematic coverage
Rapidity distributions have the weakest dependence on the value of mt
and on potential BSM contaminations (kinematic suppression of heavy resonances)
our recommendation for including 8 TeV `+jets tt̄ distributions in a fitthe normalized top rapidity distribution (1/σ)dσ/dyt from ATLAS
the normalized top-pair rapidity distribution (1/σ)dσ/dytt̄ from CMSthe corresponding total cross section σtt̄ from ATLAS and CMS
Selection minimisestension between ATLAS/CMS
tension with existing datasensitivity to mt/BSM contaminations
Balanced againstoptimal fit quality
PDF constraining powervaried kinematic coverage
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 16 / 21
Our optimal fit: features
0.7
0.8
0.9
1
1.1
1.2
1.3
0.01 0.1 1
x
g(x,Q)/gref(x,Q)
Q=100 GeV
Global baseline
Global default 0.7
0.8
0.9
1
1.1
1.2
1.3
0.01 0.1 1
x
c(x,Q)/cref(x,Q)
Q=100 GeV
Global baseline
Global default 0.7
0.8
0.9
1
1.1
1.2
1.3
0.01 0.1 1
x
b(x,Q)/bref(x,Q)
Q=100 GeV
Global baseline
Global default
AT
LA
Sdσ/dpt T
AT
LA
Sdσ/dyt
AT
LA
Sdσ/dytt̄
AT
LA
Sdσ/dm
tt̄
AT
LA
S(1/σ
)dσ/dpt T
AT
LA
S(1/σ
)dσ/dyt
AT
LA
S(1/σ
)dσ/dytt̄
AT
LA
S(1/σ
)dσ/dm
tt̄
AT
LA
Sσtt̄
CM
Sdσ/dpt T
CM
Sdσ/dyt
CM
Sdσ/dytt̄
CM
Sdσ/dm
tt̄
CM
S(1/σ
)dσ/dpt T
CM
S(1/σ
)dσ/dyt
CM
S(1/σ
)dσ/dytt̄
CM
S(1/σ
)dσ/dm
tt̄
CM
Sσtt̄
Fit opt 2.19 0.64 1.84 5.01 2.49 1.16 3.81 4.55 0.78 2.91 4.98 1.07 4.77 3.33 5.78 1.05 8.05 0.50
Overall good description of the tt̄ differential distributions included in the fitno evident signs of tension between ATLAS/CMS and with the rest of the dataset
Distributions not included in the fit not well described, except companion absolute distr.
Charm and bottom quark PDFs (generated radiatively) affected similarly as the gluon
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 17 / 21
Our optimal fit: phenomenological implications
( GeV )XM210 310
Glu
on -
Glu
on L
umin
osity
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
Baseline
+ top-quark differential
NNLO, global fits, LHC 13 TeV
( GeV )XM210 310
Glu
on -
Glu
on L
umin
osity
(R
el E
rr)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14Baseline
+ top-quark differential
+ inclusive jets
NNLO, global fits, LHC 13 TeV
Significant reduction of gg luminosity uncertainties at MX ≥ O(1) TeVe.g., at MX ∼ 2 TeV, uncertainties decrease from 13% to 5%
Impact of tt̄ differential data similar to that of jet datathough jet data analysed neglecting NNLO QCD corrections in the matrix element
A precision determination of the gluon PDF at large x is now possible at NNLOthe situation should only improve thanks to the recent NNLO jet calculation
tt̄ differential distributions are included in the new NNPDF3.1 PDF set[arXiv:1706.00428, accepted for publication in EPJC]
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 18 / 21
The gluon PDF in NNPDF3.1
x 4−10 3−10 2−10 1−10
) [r
ef]
2)
/ g (
x, Q
2g
( x,
Q
0.9
0.95
1
1.05
1.1
1.15 NNPDF3.1
NNPDF3.1 no top
NNPDF3.1 NNLO, Q = 100 GeV
x 4−10 3−10 2−10 1−10
) [r
ef]
2)
/ g (
x, Q
2g
( x,
Q
0.9
0.95
1
1.05
1.1
1.15 NNPDF3.1
NNPDF3.1 no jet
NNPDF3.1 NNLO, Q = 100 GeV
x 4−10 3−10 2−10 1−10
) [r
ef]
2)
/ g (
x, Q
2g
( x,
Q
0.9
0.95
1
1.05
1.1
1.15 NNPDF3.1
NNPDF3.1 no Z pT
NNPDF3.1 NNLO, Q = 100 GeV
x 4−10 3−10 2−10 1−10
) [r
ef]
2)
/ g (
x, Q
2g
( x,
Q
0.9
0.95
1
1.05
1.1
1.15 NNPDF3.1
NNPDF3.0
NNLO, Q = 100 GeV
Nice consistency of the three data sets in driving the gluon PDF
The central value changes by a significant amount, compatible with reduced uncertainties
NNPDF3.1 is more accurate than NNPDF3.0
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 19 / 21
Luminosities and Higgs production
( GeV )XM10 210 310
Glu
on -
Glu
on L
umin
osity
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2 NNPDF3.1
NNPDF3.0
LHC 13 TeV, NNLO
( GeV )XM10 210 310
Qua
rk -
Glu
on L
umin
osity
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2 NNPDF3.1
NNPDF3.0
LHC 13 TeV, NNLO
enhancement at small MX , suppression at large MX ; shift below one-sigmauncertainties of 1− 2% in a wide range |y| . 2 and 100 GeV.MX . 1 TeV
0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04Ratio to NNPDF3.1
ggHiggs.v3.5
LHC 13 TeVmH = 125 GeV
Higgs production: gluon fusion
NNPDF3.1
NNPDF3.0
CT14
MMHT14
ABMP16
0.80 0.85 0.90 0.95 1.00 1.05 1.10Ratio to NNPDF3.1
mg5 amc@NLO
LHC 13 TeVmH = 125 GeV
Higgs production: associate production with tt̄
NNPDF3.1
NNPDF3.0
CT14
MMHT14
ABMP16
reduced uncertainties for Higgs production in gg fusion and tt̄h
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 20 / 21
Luminosities and Higgs production
4 2 0 2 4y
101
102
103
104
MX (
GeV)
Relative uncertainty for gg-luminosityNNPDF 3.1 NNLO - s = 13000.0 GeV
1
5
10
25
50
Rela
tive
unce
rtain
ty (%
)
4 2 0 2 4y
101
102
103
104
MX (
GeV)
Relative uncertainty for gq-luminosityNNPDF 3.1 NNLO - s = 13000.0 GeV
1
5
10
25
50
Rela
tive
unce
rtain
ty (%
)
enhancement at small MX , suppression at large MX ; shift below one-sigmauncertainties of 1− 2% in a wide range |y| . 2 and 100 GeV.MX . 1 TeV
0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04Ratio to NNPDF3.1
ggHiggs.v3.5
LHC 13 TeVmH = 125 GeV
Higgs production: gluon fusion
NNPDF3.1
NNPDF3.0
CT14
MMHT14
ABMP16
0.80 0.85 0.90 0.95 1.00 1.05 1.10Ratio to NNPDF3.1
mg5 amc@NLO
LHC 13 TeVmH = 125 GeV
Higgs production: associate production with tt̄
NNPDF3.1
NNPDF3.0
CT14
MMHT14
ABMP16
reduced uncertainties for Higgs production in gg fusion and tt̄h
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 20 / 21
Summary and final remarksA new landscape
A combination of theoretical and experimental advances are shedding light on the gluonPrecise constraints on the large-x gluon PDF are now available at NNLO
from LHC top-pair differential data and inclusive jet data
Including LHC 8 TeV top-pair differential distributions in a fit
tension between ATLAS and CMS data, especially for the mtt̄ distributionstension between ATLAS/CMS data and fixed-target inclusive DIS data
however, the data shows an excellent consistency in the pull upon gluon PDFreduction in uncertainties is comparable to the constraining power of inclusive jets
inclusion of rapidity distributions minimises dependence on mt and BSM resonancestt̄ data provides PDFs for precise predictions of (B)SM processes at large MX
tt̄ data are an essential ingredient in a next generation of global PDF sets
Outlook
estimate of the correlations among different distributions and between ATLAS and CMSavailability of measurements at 13 TeV (increased statistics and kinematic range)
computation of NNLO corrections for top quark differential distributions with top decays
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 21 / 21
Summary and final remarksA new landscape
A combination of theoretical and experimental advances are shedding light on the gluonPrecise constraints on the large-x gluon PDF are now available at NNLO
from LHC top-pair differential data and inclusive jet data
Including LHC 8 TeV top-pair differential distributions in a fit
tension between ATLAS and CMS data, especially for the mtt̄ distributionstension between ATLAS/CMS data and fixed-target inclusive DIS data
however, the data shows an excellent consistency in the pull upon gluon PDFreduction in uncertainties is comparable to the constraining power of inclusive jets
inclusion of rapidity distributions minimises dependence on mt and BSM resonancestt̄ data provides PDFs for precise predictions of (B)SM processes at large MX
tt̄ data are an essential ingredient in a next generation of global PDF sets
Outlook
estimate of the correlations among different distributions and between ATLAS and CMSavailability of measurements at 13 TeV (increased statistics and kinematic range)
computation of NNLO corrections for top quark differential distributions with top decays
Thank youEmanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 21 / 21
Extra material
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 1 / 9
Data/Theory comparison: ptT
0
0.3
0.6
0.9
1.2
1.5
1.8
2.1 dσ/dptT [pb/GeV]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
0
60
100
150
200
260
320
400
500
ptT [GeV]
Ratio to NNPDF3.0
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008 (1/σ)dσ/dptT [1/GeV]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
0
60
100
150
200
260
320
400
500
ptT [GeV]
Ratio to NNPDF3.0
PDF set ATLAS dσ/dptT CMS dσ/dptT ATLAS (1/σ)dσ/dptT CMS (1/σ)dσ/dptT
NNPDF3.0 0.84 (0.66) 1.24 (0.91) 3.13 (0.94) 2.03 (0.51)HERA2.0 1.13 (1.69) 0.69 (0.84) 5.19 (1.73) 1.12 (0.33)ABM12 6.23 (1.94) 12.5 (3.00) 14.0 (4.90) 2.80 (0.80)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 2 / 9
Data/Theory comparison: mtt̄
0
0.2
0.4
0.6
0.8
1
1.2
1.4 dσ/dmtt- [pb/GeV]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
345
400
470
550
650
800
1100
1600
mtt- [GeV]
Ratio to NNPDF3.0
0
0.001
0.002
0.003
0.004
0.005
0.006 (1/σ)dσ/dmtt- [1/GeV]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
345
400
470
550
650
800
1100
1600
mtt- [GeV]
Ratio to NNPDF3.0
PDF set ATLAS dσ/dmtt̄ CMS dσ/dmtt̄ ATLAS (1/σ)dσ/dmtt̄ CMS (1/σ)dσ/dmtt̄
NNPDF3.0 0.77 (0.38) 5.73 (4.36) 1.57 (0.10) 10.6 (3.87)HERA2.0 1.40 (1.30) 3.32 (1.49) 4.36 (0.30) 5.96 (2.28)ABM12 5.72 (3.81) 5.23 (3.22) 21.1 (1.61) 1.24 (0.47)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 3 / 9
Data/Theory comparison: yt
20
40
60
80
100
120
140 dσ/dyt [pb]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
-2.5
-1.6
-1.2
-0.8
-0.4 0
0.4
0.8
1.2
1.6
2.5
yt
Ratio to NNPDF3.0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45 (1/σ)dσ/dytNNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
-2.5
-1.6
-1.2
-0.8
-0.4 0
0.4
0.8
1.2
1.6
2.5
yt
Ratio to NNPDF3.0
PDF set ATLAS dσ/dyt CMS dσ/dyt ATLAS (1/σ)dσ/dyt CMS (1/σ)dσ/dyt
NNPDF3.0 0.73 (0.28) 3.04 (1.05) 4.06 (2.85) 3.29 (1.49)HERA2.0 0.72 (0.99) 3.65 (1.49) 1.76 (1.62) 4.99 (2.29)ABM12 5.32 (1.45) 22.1 (9.78) 7.09 (15.5) 17.7 (8.72)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 4 / 9
Data/Theory comparison: ytt̄
20
40
60
80
100
120
140 dσ/dytt- [pb]
NNLO theory
NNPDF3.0HERAPDF2.0
ABM12CMS
ATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
-2.5
-1.3
-0.9
-0.6
-0.3 0
0.3
0.6
0.9
1.3
2.5
ytt-
Ratio to NNPDF3.0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55 (1/σ)dσ/dytt-
NNLO theoryNNPDF3.0
HERAPDF2.0abm12
CMSATLAS
0.7
0.8
0.9
1
1.1
1.2
1.3
-2.5
-1.3
-0.9
-0.6
-0.3 0
0.3
0.6
0.9
1.3
2.5
ytt-
Ratio to NNPDF3.0
PDF set ATLAS dσ/dytt̄ CMS dσ/dytt̄ ATLAS (1/σ)dσ/dytt̄ CMS (1/σ)dσ/dytt̄
NNPDF3.0 0.84 (0.21) 0.99 (0.74) 3.59 (1.48) 1.17 (0.75)HERA2.0 0.53 (0.74) 1.02 (0.78) 1.20 (0.60) 1.23 (0.73)ABM12 4.04 (1.05) 18.0 (5.48) 20.2 (6.06) 8.26 (4.52)
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 5 / 9
Data/Theory comparison: σtt̄
0.8
0.9
1
1.1
1.2
CMS 7 TeV CMS 8 TeV CMS 13 TeV ATLAS 7 TeV ATLAS 8 TeV ATLAS 13 TeV
Ratio to NNPDF3.0 (σtt-)
NNLO theory
exp. dataNNPDF3.0
MMHT14CT14
HERAPDF2.0ABM12
PDF set ATLAS σtt̄ CMS σtt̄
NNPDF3.0 2.21 1.36MMHT14 2.51 1.32CT14 2.19 1.72HERA2.0 0.74 0.21ABM12 12.6 16.7
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 6 / 9
Sensitivity of tt̄ absolute distributions to mt (NLO)
0
1
2
3
4
5
dσ
NL
O/dp
T,a
vt
[pb/G
eV
]
mt = 172.3 GeV
mt = 173.3 GeV
0 200 400 600 800 1000
pT,avt [GeV]
1.00
1.01
1.02
1.03
1.04
1.05
rati
o
0.5
1.0
1.5
2.0
2.5
3.0
3.5
dσ
NL
O/dmtt̄
[pb/G
eV
]
mt = 172.3 GeV
mt = 173.3 GeV
400 500 600 700 800 900 1000
mtt̄ [GeV]
1.00
1.01
1.02
1.03
1.04
1.05
rati
o
−2 −1 0 1 2yavt
0.998
1.000
1.002
1.004
1.006
1.008
mass
sen
siti
vit
y p p→ tt̄ (8 TeV)µF = µR = HT/4NNPDF3.0
LO NLO
−2 −1 0 1 2ytt̄
0.998
1.000
1.002
1.004
1.006
1.008
mass
sen
siti
vit
y p p→ tt̄ (8 TeV)µF = µR = HT/4NNPDF3.0
LO NLO
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 7 / 9
Phenomenological implications
( GeV )XM210 310
Glu
on -
Glu
on L
umin
osity
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
Baseline
+ top-quark differential
NNLO, global fits, LHC 13 TeV
( GeV )XM210 310
Glu
on -
Glu
on L
umin
osity
(R
elE
rr)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Baseline
+ top-quark differential
NNLO, global fits, LHC 13 TeV
( GeV )XM210 310
Qua
rk -
Ant
iqua
rk L
umin
osity
0.85
0.9
0.95
1
1.05
1.1
1.15
Baseline
+ top-quark differential
NNLO, global fits, LHC 13 TeV
( GeV )XM210 310
Qua
rk -
Ant
iqua
rk L
umin
osity
(R
elE
rr)
0
0.02
0.04
0.06
0.08
0.1
Baseline
+ top-quark differential
NNLO, global fits, LHC 13 TeV
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 8 / 9
Phenomenological implications
0
0.3
0.6
0.9
1.2
1.5
1.8
2.1 dσ/dptT [pb/GeV]
NNLO theory
LHC 8 TeV
NNPDF (before)NNPDF (after)
CMSATLAS
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
1.35
0
60
100
150
200
260
320
400
500
ptT [GeV]
Ratio to baseline
0
0.2
0.4
0.6
0.8
1
1.2
1.4 dσ/dmtt- [pb/GeV]
NNLO theory
LHC 8 TeV
NNPDF (before)NNPDF (after)
CMSATLAS
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
345
400
470
550
650
800
1100
1600
mtt- [GeV]
Ratio to baseline
Emanuele R. Nocera (Oxford) The large-x gluon with tt̄ differential data 6th September 2017 9 / 9