PoWHEG + Pythia matching
James MonkKiran Joshi
Lily AsquithFelix Müller
Introduction
17
๏ In extreme cases, the shower may be allowed to run wild, e.g. this old version of PoWHEG that has since been patched
๏ PoWHEG provides a scale (SCALUP) that is an indication of where the shower should take over from the perturbative calculation.
๏ There may be some mis-matching between Pwg and Pythia about what is considered the scale of an emission. e.g. pT isn’t very indicative of the hardness of a very forward emission.
partonT
p0 20 40 60 80 100 120 140 160 180 200
truth
Tp
0
20406080
100120140
160180200
#jet
s
1
10
210
310
410
510 = 2.76 TeVsPythia/AUET2bno weights
=5GeVmintborn k
J4
J1
J2
J3
Alternative matching schemes
18
Leg-emission veto
5
For each proposed shower emission, ps, calculate a scale μs
4
1
2
35
µs = maxi
|ps × pj ||pj |
PoWHEG legs
Start the shower at the kinematic limit (“power shower”), but veto any
emission for which μs is above the (unchanged) PoWHEG veto scale
PoWHEG Scale Recalculation Scheme
Recalculate the veto scale, μ, based on the configuration of the PoWHEG legs
4
4
1
2
35
µ = mini,j
|pi × pj ||pj |
Start the shower at the kinematic limit (“power
shower”), but veto any emission above this calculated μ
Original Scale [GeV]0 50 100 150 200 250 300 350 400 450 500
New
Sca
le [G
eV]
0
50
100
150
200
250
300
350
400
450
500
1
10
210
310
Old Vs new scales
PoWHEG legs
Suggested by PoWHEG authors as a way to explore systematics
of (reducing) scale choice
Suggested by Pythia authors as a way to better measure how hard an emission is
Improved ISR description!..
19
Multi-jets
11
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalculationLeg-emission veto
10!1
Inclusive Jet Multiplicity Ratio (R=0.4)
!N
/!
N!
1
3 4 5 6
0.6
0.8
1
1.2
1.4
Njet
MC
/D
ata
Note in particular the 3:2 ratio, which you’d hope to get right with the PoWHEG emission
(something odd in the 4:3 ratio though)
Improved ISR description!..
20
Azimuthal decorrelations
16
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalculationLeg-emission veto
10!2
10!1
1
Dijet azimuthal decorrelations for 160 < pmax"
/GeV < 210
1/!
d!
/d
!"
[#/
rad
]
0.5 0.6 0.7 0.8 0.9 1.00
0.5
1
1.5
2
!" [rad/#]
MC
/D
ata
A clear win for the new matchings over the vanilla!
leg-emission matching may be slightly better, but scale reduction is close
All distributions show similar
…but (fatally) worsened FSR description
21
Jet shapes
17
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalculationLeg-emission veto
1
Jet shape ! for p! " 40–60 GeV, y " 1.2–2.1
!(r
)
0 0.1 0.2 0.3 0.4 0.5 0.6
0.6
0.8
1
1.2
1.4
r
MC
/D
ata
The new matchings have broken the jet shapes though!
The fact that the jet shapes is so good in the vanilla is a bit surprising - not a combined PoWHEG + Pythia tune
New matching gives too much of a hard core to the jet
Bolder variationTurn renormMultFac down to 0.5:
Low pT: High pT:
→ ρ is pretty insensitive even to large variation.
Look at other |y | bins...
5/9
Varying e.g. alpha_s in shower suggested it would not be
possible to re-obtain the good jet shape behaviour
Ad Hoc Solution: only apply the modified matching to ISR
22
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalcLeg-emission vetoScale recalc, ISR onlymain31, emt010!1
Inclusive jet multiplicity ratio N/N ! 1 (R = 0.4)
!N
/!
N!
1
3 4 5 6
0.6
0.8
1
1.2
1.4
Njet
MC
/D
ata
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalcLeg-emission vetoScale recalc, ISR onlymain31, emt0
1
Jet shape ! for p! " 40–60 GeV, y " 0.8–1.2
!(r
)
0 0.1 0.2 0.3 0.4 0.5 0.6
0.6
0.8
1
1.2
1.4
r
MC
/D
ata
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalcLeg-emission vetoScale recalc, ISR onlymain31, emt0
10!2
10!1
1
Dijet azimuthal decorrelations for 160 < pmax"
/GeV < 210
1/!
d!
/d
!"
[#/
rad
]
0.5 0.6 0.7 0.8 0.9 1.00
0.5
1
1.5
2
!" [rad/#]
MC
/D
ata
Improves many distributions while keeping the jet shapes unchanged.
ATLAS datavanilla Py8 AU2-CTEQ6L1
Pwg scale recalcLeg-emission vetoScale recalc, ISR onlymain31, emt0
10!4
10!3
10!2
10!1
1
10 1
10 2
10 3
10 4
10 5
10 6
10 7Incl. jet double-diff. x-section, anti-kt 0.4 (|y| < 0.3)
d2!
/d
p "d
y[p
b/
GeV
]
200 400 600 800 1000 1200 1400
0.6
0.8
1
1.2
1.4
p" [GeV]
MC
/D
ata
Even slightly improves differential cross section
at low pT
“Main31”
23
๏ Example matching provided with Pythia. Initially showed very poor description, but recently fixed after interactions with authors.
๏ Author recommended setup. Applies to both ISR and FSR (although agree that there is more theoretical freedom to vary FSR scale choice)
๏ Performs almost as well as ISR-only scale reduction (and much better than vanilla setups). Since it’s less ad-hoc, our likely recommendation for future QCD production
๏ ISR-only setup can provide a nice systematic variation - they both perform well on these distributions.
๏ Full set of plots here: http://www.nbi.dk/~jmonk/powhegmatching/
Top, Z
24
ATLAS dataMC (main31)MC (PowhegPythia6 P2011C)
0.7
0.75
0.8
0.85
0.9
0.95
1.0
Gap fraction vs. Q0 for veto region: |y| < 0.8
f gap
50 100 150 200 250 300
0.96
0.98
1.0
1.02
1.04
Q0 [GeV]
MC/data
ATLAS dataMC (main31)MC (PowhegPythia6 P2011C)
0.5
0.6
0.7
0.8
0.9
1.0
Gap fraction vs. Q0 for veto region: |y| < 2.1
f gap
50 100 150 200 250 300
0.9
0.95
1.0
1.05
1.1
Q0 [GeV]
MC/data
Very preliminary studies of top veto by Kiran Joshi suggest there may be issues, but the picture isn’t clear yet
ATLAS dataVetoed showerWimpy showerDefault shower
10!6
10!5
10!4
10!3
10!2
10!1Z p" reconstructed from dressed muons
1/!
d!
/d
p"
[GeV
!1]
1 10 1 10 2
0.6
0.8
1
1.2
1.4
p"(µµ) [GeV]
MC
/D
ata
Did not affect the Z pT distribution much, but
maybe W/Z + jets (c.f. jet Δϕ) will show a
difference. No FSR ambiguity in this case