uncertainties in predicting ttbb by powhel · 9 choice of scales ‣ qcd corrections are ‣ large...
TRANSCRIPT
Zoltán Trócsányi !
University of Debrecen and MTA-DE Particle Physics Research Group in collaboration with
A. Kardos, M.V. Garzelli !!!!!!
Higss Cross Section Working Group November 10, 2014
Uncertainties in predicting ttbb by PowHel
Introduction
3
possible theoretical predictions: !
tt at NLO + b(b)-jet by SMC (POWHEG or MadGraph5_aMC@NLO) tt + jet (can be b-jet) at NLO + b(b) by SMC (PowHel) tt + 2jets (one can be b-jet) at NLO + b(b) by SMC (SHERPA+OpenLoops) tt + bb at NLO + SMC: in this talk
modeling tt+HF jets at NLO+PS
4
missing higher orders: QCD: largest, but NNLO is not feasible during run 2 ⇒ estimated by variation of renormalization and
factorization scales, µR = µF = µ0 in [µ0/2, 2 µ0]
EW: NLO not known, but expected to be small except perhaps for large transverse momenta of t-quarks or jets PDF: take envelope of predictions made using various PDF sets neglected b-quark mass: <3% at LO [1001.4006] treatment of t → b l νl decay if included, NLO for
pp → b l νl + b l νl + bb is not available at present
Uncertainties in NLO predictions
5
two methods of matching: MC@NLO: SHERPA+OpenLoops (phenomenology in arXiv:1309.5912 and next talk) MadGraph5_aMC@NLO (phenomenology in third talk) POWHEG: PowHel (phenomenology in J. Phys. G 41 (2014) 075005 [arXiv:1303.6291], arXiv:1307.1347 and 1408.0266, also in this talk)
Matching NLO to PS
6
matching uncertainty choice of SMC and its tune neglected truncated showers: likely negligible, but not checked approximate treatment of t → b l νl decay (our
option: DECAYER) !
here we study scale uncertainty at NLO matching uncertainty SMC uncertainty PDF and scale uncertainties in NLO+PS
focusing on hardest b-jets
Uncertainties in PowHel
scale uncertainties at NLO
8
Choice of scales
‣ QCD corrections are ‣ large with scales µ0 = mt or mt+mbb/2 (about 80%) ‣ moderate with dynamical scale µ0=(mt2 pT,bpT,b)1/4
(about 25%) (proposed in arXiv:1001.4006), implying better convergence by emulating higher order effects through CKKW-type scale choice
−
−
9
Choice of scales
‣ QCD corrections are ‣ large with scales µfix = mt or mt+mbb/2 (about 70%) ‣ moderate with dynamical scale µdyn= (mt2 pT,bpT,b)1/4
(about 25%) (proposed in arXiv:1001.4006), implying better convergence by emulating higher order effects through CKKW-type scale choice,
but
‣ we want to simulate higher order effects through the PS: µdyn is too small near threshold where cross section is largest, even for a b with pT = 100 GeV and another b with pT = 20 GeV µdyn = 90 GeV << mt resulting in an artificially large xsection at LO
−
−
10
Choice of scales
We use the dynamical scale µdyn= HT/2, where HT is the scalar sum of transverse masses of final-state particles that is a good scale also near threshold
With this scale
✓ the K factor is even smaller, implying good convergence
✓ the cross sections are
smaller (w/ cuts of 1001.4006):
σLO = 534 fb, σNLO = 630 fb, K = 1.18
scale dependence: +32%-22%, largest if µR = µF = µdyn
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
�[pb]
(i) NLOLO
0.8
1.0
1.2
K-factor
10
�12 5 1 2
2µ/H?
11
Small changes in shapes of distributions
2
5
1
2
5
10
2d�
dp ?
,b1[fb/G
eV]
(a) NLO: µ = H?/2,LO: µ = H?/2,
H?/4 µ H?H?/4 µ H?
1.0
1.3
K-factor
50 100 150 200 250 300 350 400
p?,b1 [GeV]
2
5
1
2
5
10
2
5
d�
dp ?
,b2[fb/G
eV]
(b) NLO: µ = H?/2,LO: µ = H?/2,
H?/4 µ H?H?/4 µ H?
1.0
1.3
K-factor
20 40 60 80 100 120 140 160 180 200
p?,b2 [GeV]
0
50
100
150
200
250
300
350
400
d�
dy b1[fb]
(a) NLO: µ = H?/2,LO: µ = H?/2,
H?/4 µ H?H?/4 µ H?
1.0
1.3
K-factor
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
yb1
0
50
100
150
200
250
300
350
d�
dy b2[fb]
(b) NLO: µ = H?/2,LO: µ = H?/2,
H?/4 µ H?H?/4 µ H?
1.0
1.3
K-factor
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
yb2
12
Small changes in shapes of distributions
2
5
1
2
5
10
2
d�
dm
bb̄[fb/G
eV]
(b) NLO: µ = H?/2,LO: µ = H?/2,
H?/4 µ H?H?/4 µ H?
1.0
1.3
K-factor
100 150 200 250 300 350 400
mbb̄ [GeV]
matching uncertainty
14
Formal accuracy of the POWHEG MC
hOi =Z
d⇥BeB�(p? ,min)O(⇥B) +
Zd⇥rad�(p?)
R
BO(⇥R)
�=
=
⇢Zd�B [B + V ]O(�B) +
Zd�RRO(�R)
�(1 +O(�S))
�O⇥NLOUseful for checking
...
15
LHE vs. NLO
2
5
1
2
5
10
2
5
d�
dp ?
,b1[fb
/GeV
]
(a) NLO: µ = H?/2,LHE: µ = H?/2,
H?/4 µ H?H?/4 µ H?
0.81.01.2
NLO
LHE
50 100 150 200 250 300 350 400
p?,b1 [GeV]
2
5
1
2
5
10
2
5
d�
dp ?
,b2[fb
/GeV
]
(a) NLO: µ = H?/2,LHE: µ = H?/2,
H?/4 µ H?H?/4 µ H?
0.81.01.2
NLO
LHE
20 40 60 80 100 120 140 160 180 200
p?,b2 [GeV]
0
50
100
150
200
250
300
350
400
d�
dy b1[fb
]
(b) NLO: µ = H?/2,LHE: µ = H?/2,
H?/4 µ H?H?/4 µ H?
0.81.01.2
NLO
LHE
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb1
0
50
100
150
200
250
300
350
400
d�
dy b2[fb
]
(b) NLO: µ = H?/2,LHE: µ = H?/2,
H?/4 µ H?H?/4 µ H?
0.81.01.2
NLO
LHE
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb2
16
LHE: distributions from events at Born+1st radiation
Decay: on-shell decays of heavy particles (t-quarks), shower and hadronization effects turned off
PS: parton showering (PYTHIA or HERWIG) included (t-quarks kept stable)
Full SMC: decays, parton showering and hadronization are included by using PYTHIA or HERWIG
Number and type of particles are very different => to check that SMC does what we expect, we employ selection cuts to keep the cross section fixed
Four possible forms of predictions
17
Selection cuts for decay vs. SMC
‣ Applied on the LHE’s: ‣ A track was considered as a possible jet
constituent if |ηtrack|<5, t-quarks were excluded from the set of possible tracks. Jets were reconstructed with the anti-kT algorithm using R=0.4.
‣ Events with invariant mass of the bb-jet pair below mminbb = 100 GeV were discarded.
‣ Applied on LHE’s and checked also on the existing particles at different stages of evolution:
‣ we require pTmin,j = 25 GeV and
‣ at least two, one b- & one b-jet with |ηb(b)|<2.5.
−
− −
−
18
NLO vs. PS vs. LHE at 14TeV, µ = HT/2
10�2
2
5
10�1
2
5
12
5
10
d�
dp ?
,b1[fb
/GeV
]
NLO PSLHE
0.8
1.0
NLO,PS
LHE
1.0
1.2
PS
NLO
0 50 100 150 200 250 300 350 400
p?,b1 [GeV]
10�1
2
5
1
2
5
10
2
d�
dp ?
,b2[fb
/GeV
]
(b) NLO PSLHE
0.8
1.0
NLO,PS
LHE
0.8
1.0
PS
NLO
0 20 40 60 80 100 120 140 160 180 200
p?,b2 [GeV]
19
NLO vs. PS vs. LHE at 14TeV, µ = HT/2
10
2
5
102
2
5
103
d�
dy b1[fb
]
NLO PSLHE
0.8
1.0
NLO,PS
LHE
1.0
1.2
PS
NLO
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb1
10
2
5
102
2
5
103
d�
dy b2[fb
]
(b) NLO PSLHE
0.8
1.0
NLO,PS
LHE
1.0
1.2
PS
NLO
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb2
20
NLO vs. PS and decay vs. full SMC at 14TeV, µ = HT/2
10�1
2
5
1
2
5
10
d�
dm
bb̄[fb
/GeV
]SMCNLO
decayPSLHE
0.60.81.0
NLO,PS
LHE
0.81.01.2
decay
SMC
0 50 100 150 200 250 300 350 400
mbb̄ [GeV]
21
Message: matching and SMC is under control
decay of t-quarks can have big impact
22
Cuts for estimating the effect of the PS
applied separately on LHEs and after PS
‣ jets reconstructed with the anti-kT algorithm using R=0.5, with at least two well-separated b-jets (ΔR > 0.5), pTmin,bjet = 40 GeV and |ηj|<2.5
but with
‣ t-quarks kept stable
23
PS vs. LHE at 14TeV, µ = HT/2
10�1
2
5
1
2
5
10
2
d�
dp ?
,b1[fb/G
eV]
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
1.01.11.2
PS/L
HE
0 50 100 150 200 250 300 350 400
p?,b1 [GeV]
5
102
2
5
103
d�
dy b1[fb]
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
1.01.11.2
PS/L
HE
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb1
10�1
2
5
1
2
5
10
2
5
102
d�
dp ?
,b2[fb/G
eV]
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
1.01.11.2
PS/L
HE
0 20 40 60 80 100 120 140 160 180 200
p?,b2 [GeV]
5
102
2
5
103
d�
dy b2[fb]
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
1.01.11.2
PS/L
HE
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb2
24
PS vs. LHE at 14TeV, µ = HT/2
10
�7
10
�6
10
�5
10
�4
10
�3
10
�2
10
�1
1
fractionofN-jets
2 4 6 8 10 12 14 16 18 20
N
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
CT10, tt̄+PYTHIA-6
10�1
2
5
1
2
5
10
d�
dm
bb̄[fb/G
eV]
CT10, LHE
CT10, PYTHIA-6
CT10, PYTHIA-8
1.01.11.2
PS/L
HE
0 50 100 150 200 250 300 350 400
mbb̄ [GeV]
25
Message: PYTHIA-6 and PYTHIA-8 give similar predictions but this may depend on selection cuts (see below)
PDF and scale uncertainties of NLO+PS predictions
27
Cuts for scale, PDF and SMC uncertainties
from CMS PAS TOP-13-010
‣ jets reconstructed with the anti-kT algorithm using R=0.5, with pTmin,j = 40 GeV and |ηj|<2.5
‣ at least one pair of isolated (with R=0.3, Irel = 0.15) opposite sign leptons with pTmin,l = 20 GeV/c, |ηl|<2.4, 12 GeV < mllc2 (∉[77, 107] GeV if ee or µµ)
‣ pTmiss = 30 GeV/c if ee or µµ
‣ at least four well separated b-jets with ΔR > 0.5 both from leptons and jets
28
PDF and scale uncertainties
10�3
2
5
10�2
2
5
10�1
2
d�
dp ?
,b1
[fb/G
eV]
(a)
CT10MSTWNNPDF
µ 2 [H?/4, H?]PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
0 50 100 150 200 250 300 350 400
p?,b1
[GeV]
10�3
2
5
10�2
2
5
10�1
2
d�
dp ?
,b2
[fb/G
eV]
(a)
CT10MSTWNNPDF
µ 2 [H?/4, H?]PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
0 50 100 150 200 250 300 350 400
p?,b2
[GeV]
29
PDF and scale uncertainties
10�2
2
5
10�1
2
5
12
5
10
d�
dy b1
[fb]
(b)
CT10MSTWNNPDF
µ 2 [H?/4, H?]PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb
1
10�2
2
5
10�1
2
5
12
5
10
d�
dy b2
[fb]
(b)
CT10MSTWNNPDF
µ 2 [H?/4, H?]PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5yb
2
30
PDF and scale uncertainties
10�5
10�4
10�3
10�2
10�1
d�
dp ?
,`1
[fb/G
eV]
(a)
CT10MSTWNNPDF
µ 2 [H?/4, H?]PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
0 50 100 150 200 250 300 350 400
p?,`1
[GeV]
5
10�3
2
5
10�2
2
5
10�1
d�
dm
b¯ b[fb
/GeV
]
(a)
CT10MSTWNNPDF
µ 2 [H?/4, H?]
PDF 2 {CT10, MSTW, NNPDF}
0.71.01.3
ratio
(14TeV)
0.71.01.3
ratio
(8TeV)
0 50 100 150 200 250 300 350 400
mb¯b [GeV]
SMC uncertainties
32
PYTHIA-6 vs. PYTHIA-8 at 14TeV, µ = HT/2
10
�7
10
�6
10
�5
10
�4
10
�3
10
�2
fractionofN
jets
2 4 6 8 10 12
N
CT10, PYTHIA-8
CT10, PYTHIA-6
10
�6
10
�5
10
�4
10
�3
10
�2
fractionofN
b-jets
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
N
CT10, PYTHIA-8
CT10, PYTHIA-6
33
PYTHIA-6 vs. PYTHIA-8 at 14TeV, µ = HT/2
2
5
10�2
2
5
10�1
d�
dp ?
,b1[fb
/GeV
]
CT10, PYTHIA-8
CT10, PYTHIA-6
0.81.01.2
PY-6
PY-8
0 50 100 150 200 250 300 350 400
p?,b1 [GeV]
5
10�2
2
5
10�1
2
d�
dp ?
,b2[fb
/GeV
]
CT10, PYTHIA-8
CT10, PYTHIA-6
0.81.01.2
PY-6
PY-8
0 20 40 60 80 100 120 140 160 180 200
p?,b2 [GeV]
2
5
1
2
5
10
d�
dy b1[fb
]
CT10, PYTHIA-8
CT10, PYTHIA-6
0.81.01.2
PY-6
PY-8
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
dyb1
5
1
2
5
10
d�
dy b1[fb
]
CT10, PYTHIA-8
CT10, PYTHIA-6
0.81.01.2
PY-6
PY-8
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
dyb2
Conclusions and outlook
35
K-factor and scale uncertainty of NLO prediction is moderate with dynamical scales, but choice of central scale matters (we prefer HT/2 for NLO+PS) Matching is under control within scale uncertainty For NLO predictions matched with SMC
scale uncertainties are +35%-30% PDF uncertainties are ~ 10% SMC uncertainties are ~< 10%
…but all conclusions are sensitive to selection cuts effects of decay of t-quarks could be important
Conclusions
36
We are open to comparison of predictions from PowHel, SHERPA+OpenLoops & MadGraph5_aMC@NLO
with same set of parameters and cuts
(to be agreed together with experimentalists)
Outlook
Appendix
38
Cuts employed by Bevilacqua et al in arXiv:0907.4723
‣ A track was considered as a possible jet constituent if |ηtrack|<5, t-quarks were excluded from the set of possible tracks, jets were reconstructed with the kT-algorithm using R=0.4
‣ Events with invariant mass of the bb-jet pair below mminbb = 20 GeV were discarded
‣ We require pTmin,j = 20 GeV and
‣ at least two, one b- and one b-jet, with |yb(b)|<2.5
−
−−
−
Selection cuts for NLO predictions
39
Cuts for background study for tTH
Applied after full SMC ‣ a track was considered as a possible jet constituent if
|ηtrack|<5, jets were reconstructed with the anti-kT algorithm using R=0.4
we require
‣ at least six jets with pTmin,j = 20 GeV and |ηj|<5
‣ at least two b-jets & two b-jets with |ηb(b)|<2.7, with MCTRUTH tagging
‣ at least one isolated (with R=0.4) lepton with pTmin,l = 20 GeV and |ηl|<2.5
‣ pTmiss = 15 GeV to disentangle background in the semileptonic tt decay
− −
−
40
ttH signal on ttbb background−−−
2
5
10�3
2
5
10�2
2
5
10�1
2
5
12
5d�
dm
b 1b 2[fb
/GeV
]
50 100 150 200 250 300 350 400
mb1 b2 [GeV]
ps = 8TeV
µtt̄bb̄0 = H?/2
µtt̄H0 = mt +mH/2
mt = 173.2GeVmH = 125GeV
(iii)µ0/2 µtt̄bb̄ 2µ0
2
5
10�1
2
5
12
5
102
5
1022
5
d�
d�R(b1,b2)[fb
]
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
�R(b1 , b2)
ps = 8TeV
µtt̄bb̄0 = H?/2
µtt̄H0 = mt +mH/2
mt = 173.2GeVmH = 125GeV
(iv)µ0/2 µtt̄bb̄ 2µ0
2
5
10�3
2
5
10�2
2
5
10�1
2
5
12
5
d�
dp ?
,b1[fb
/GeV
]
50 100 150 200 250 300 350 400
p? ,b1 [GeV]
ps = 8TeV
µtt̄bb̄0 = H?/2
µtt̄H0 = mt +mH/2
mt = 173.2GeVmH = 125GeV
(i)µ0/2 µtt̄bb̄ 2µ0
2
5
10�3
2
5
10�2
2
5
10�1
2
5
12
5
d�
dp ?
,b2[fb
/GeV
]
50 100 150 200 250 300 350
p? ,b2 [GeV]
ps = 8TeV
µtt̄bb̄0 = H?/2
µtt̄H0 = mt +mH/2
mt = 173.2GeVmH = 125GeV
(ii)µ0/2 µtt̄bb̄ 2µ0