alpgen + [pythia parameter variation effects] · pythia: 6.4.23, using the new ps/mi for all the...
TRANSCRIPT
ALPGEN + [PYTHIA Parameter Variation Effects]
Ben Cooper, Judith Katzy, Michelangelo Mangano, Andrea Messina, Liza MijovicThanks to Peter Skands for useful discussions.
10. September 2010
1 / 26
Outline
Expanding on Alpgen + Pythia features reported/discussed during Ben’s talkyesterday.
1) Why is Perugia Soft + ALPGEN harder than Perugia Hard + ALPGEN?Which of the Pythia parameters (don’t) contribute to the observed difference?Sketch the main features of PYTHIA parameter variation and ALPGEN matchinginterplay.
Since it turns out PS-related parameters variation is of most relevance for the point 1):
2) Explore ISR and FSR PYTHIA parameter variation effects;Use Perugia 2010 as a central tune and increase/decrease ISR (P2010 ISR ↑, ISR ↓).Use Perugia 2010 as a central tune and increase/decrease FSR (P2010 FSR ↑, FSR ↓).Show the effects on jet-related distributions, x-sections (matching efficiency).
Hard process: V+jets, Pythia 6.4.23, pt-ordered shower, Alpgen: v2.13.
Will not address ALPGEN parameter variations.
2 / 26
Technical Information
hard process: W+ up to 4 jets @ LHC, 7 TeV
ALPGEN: 2.13ebeam 3500.
ih2 1
ickkw 1.
ptjmin 20.
drjmin 0.4
etajmax 3.0
njets: [0,1,2,3,4(inclusive)]
In addition: ET(CLUS)=ptjmin+5 GeV, R(CLUS)=drjmin, ETACLMAX=etajmax.
Pythia: 6.4.23, using the new PS/MI for all the plots.Pythia tunes (P. Skands, arXiv:1005.3457v2):Perugia soft (Psoft), Perugia hard (Phard), Perugia 0 (P0),Perugia 2010,relevant parameter variations will be explained on the fly.
Analysis:jets are obtained from generator-level stable particles using Fastjet (2.4.2)anti-kt alg. with R=0.4.Jet cuts :pt>20 GeV (or higher), and |η| <2.5.Normalization (when applied):
RL = 100 pb−1.
3 / 26
Psoft vs Phard Tune - 1
P0 = a central tune, incl. LEP Z, Tevatron MinBias and DY data.
Phard: more hard pertrubative, less soft non-perturbative activity than P0.
Psoft: less hard pertrubative, more soft non-perturbative activity than P0.
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710Psoft tune
P0 tune
Phard tune
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P0
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710Psoft tune
P0 tune
Phard tune
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.5
4 / 26
Psoft vs Phard Tune - 2
P0 = a central tune, incl. LEP Z, Tevatron MinBias and DY data.
Phard: more hard pertrubative, less soft non-pertorbative activity than P0.
Psoft: less hard pertrubative, more soft non-pertorbative activity than P0.
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710Psoft tune
P0 tune
Phard tune
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.550 100 150 200 250 300
N (
Log
scal
e)
210
310
410
510
610 Psoft tune
P0 tune
Phard tune
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250 300
dist
./P0
0.5
1
1.5
5 / 26
Psoft vs Phard Tune : hard ISR Parameters - 1
Producing 2 additional samples with P0 + (only) ISR parameters varied on top asin Phard and Psoft:
PARP(67): multiplicative factor of a hard scale that sets the allowed phase space forthe hardest ISR emission.PARP(64): multiplicative factor of the ISR evolution scale (Q2, p2
t ).MSTP(64): ISR λ is λMS (for MSTP(64)=2) or λCMW (for MSTP(64)=2).
P0: PARP(67)=1.0, PARP(64)=1.0, MSTP(64)=3,Phard: PARP(67)=4.0, PARP(64)=0.25, MSTP(64)=3,Psoft: PARP(67)=0.5, PARP(64)=2.0, MSTP(64)=2.
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710ISR(Psoft) sample
P0 tune
ISR(Phard) sample
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P0
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710ISR(Psoft) sample
P0 tune
ISR(Phard) sample
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.5
6 / 26
Psoft vs Phard Tune : hard ISR Parameters - 2
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710ISR(Psoft) sample
P0 tune
ISR(Phard) sample
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.550 100 150 200 250 300
N (
Log
scal
e)
210
310
410
510
610
710 ISR(Psoft) sample
P0 tune
ISR(Phard) sample
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250 300
dist
./P0
0.5
1
1.5
effect on high-pt jet #, pt, observed, but not as large as in Psoft, Phard,
little (wrt. Psoft, Phard diff.) effect on pt>20 GeV jets # distribution; thedifference for this distribution comes from other sources (identified in the nextslides).
7 / 26
Psoft vs Phard Tune : IFSR Parameters - 1Producing 2 additional samples with P0 + (only) IFSR parameters varied on topas in Phard and Psoft:
PARP(71): multiplicative factor of a hard scale that sets the allowed phase space forthe hardest FSR emission.MSTP(72): switch setting a detail of the FSR off ISR (have pasted the definition to thelast slide).n.b.: none of these is the FSR λQCD .
P0: PARP(71)=2.0, MSTP(72)=1,Phard: PARP(71)=4.0, MSTP(72)=1,Psoft: PARP(71)=1.0, MSTP(72)=0 (less FSR off ISR).
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710IFSR(Psoft) sample
P0 tune
IFSR(Phard) sample
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P0
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710IFSR(Psoft) sample
P0 tune
IFSR(Phard) sample
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.5
8 / 26
Psoft vs Phard Tune : IFSR Parameters - 2
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710IFSR(Psoft) sample
P0 tune
IFSR(Phard) sample
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.550 100 150 200 250 300
N (
Log
scal
e)
210
310
410
510
610
710 IFSR(Psoft) sample
P0 tune
IFSR(Phard) sample
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250 300
dist
./P0
0.5
1
1.5
Apart from ISR also the IFSR variations also contribute to jet (pt>30, 40) #differences between Psoft and Phard.
9 / 26
Psoft vs Phard Tune : UE Parameters - 1
Producing 2 additional samples with P0 + (only) IFSR parameters varied on topas in Phard and Psoft:
PARP(82): MI IR pt cutoff,
MSTP(90): energy scaling of the MI IR pt cutoff (as EPARP(90)CM ),
PARP(83): hadronic matter overlap,net effect is more non-perturbative activity in Psoft and less in Phard (wrt. P0).
P0: PARP(82)=2.0, PARP(90)=0.26, PARP(83)=1.7,Phard: PARP(82)=2.3, PARP(90)=0.30, PARP(83)=1.7,Psoft: PARP(82)=1.9, PARP(90)=0.24, PARP(83)=1.5.
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710UE(Psoft) sample
P0 tune
UE(Phard) sample
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P0
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710UE(Psoft) sample
P0 tune
UE(Phard) sample
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.5
10 / 26
Psoft vs Phard tune : UE Parameters - 2
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710UE(Psoft) sample
P0 tune
UE(Phard) sample
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P0
0.5
1
1.550 100 150 200 250 300
N (
Log
scal
e)
210
310
410
510
610
710 UE(Psoft) sample
P0 tune
UE(Phard) sample
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250 300
dist
./P0
0.5
1
1.5
effect on pt>20 GeV # is as expected and also observed when using Pythia stand.
effect on jet # is negligible (wrt. e.g. ISR effects) for pt>40(30) GeV jets.
11 / 26
Effects on x-Section (Matching Efficiency)
Psoft and Phard values differ from P0 ones for ∼ 10-20%;
this can be assigned to ISR and IFSR differences between Psoft and Phard,
it is not caused by Underlying Event (UE), Colour Reconnections (CR) or othernon-perturbative differences between the tunes.
tune 0j 1j 2j 3j 4j total [pb]Phard 7287 ± 3.9 728 ± 2.6 141 ± 1.3 27 ± 0.2 6.6 ± 0.2 8190 ± 8P0 7556± 3.6 814 ± 2.7 166 ± 1.3 32 ± 0.3 7.8 ± 0.3 8576 ± 8Psoft 7804 ± 3.4 944 ± 2.8 207 ± 1.5 42 ± 0.3 10.1± 0.3 9007 ± 8P0 ISR (Phard) 7207 ± 6.9 735 ± 2.6 143 ± 1.3 27 ± 0.2 6.9 ± 0.2 8119 ± 11P0 ISR (Psoft) 7831 ± 4.9 881 ± 2.7 186 ± 1.4 36 ± 0.3 8.8 ± 0.3 8943 ± 10P0 FISR (Phard) 7548 ± 6.0 814 ± 2.7 167 ± 1.3 32 ± 0.3 7.8 ± 0.3 8569 ± 10P0 FISR (Psoft) 7505 ± 6.1 878 ± 2.7 188 ± 1.4 37 ± 0.3 9.4 ± 0.3 8617 ± 10P0 UE (Phard) 7513 ± 6.1 826 ± 2.7 171 ± 1.4 33 ± 0.3 7.8 ± 0.3 8551 ± 10P0 UE (Psoft) 7576 ± 5.9 817± 2.7 166 ± 1.3 32 ± 0.3 8.1 ± 0.3 8599 ± 10P0 CR (Phard) 7561 ± 5.9 821 ± 2.7 167 ± 1.3 32 ± 0.3 8.1 ± 0.3 8589 ± 10P0 CR (Psoft) 7556 ± 5.9 815 ± 2.7 165 ± 1.3 32 ± 0.3 8.1 ± 0.3 8576 ± 10
12 / 26
P2010 Tune + ISR Parameters Variation - 1
Producing 2 additional samples with P2010 + (only) ISR parameters varied ontop:
PARP(67): multiplicative factor of a hard scale that sets the allowed phase space forthe hardest ISR emission.PARP(64): multiplicative factor of the ISR evolution scale (Q2, p2
t ).
P2010: PARP(67)=1.0, PARP(64)=1.0,P2010 ISR ↑: PARP(67)=4.0, PARP(64)=0.25,P2010 ISR ↓: PARP(67)=0.5, PARP(64)=4.0.
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 ISR
P2010 tune
↑P2010 ISR
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P20
10
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 ISR
P2010 tune
↑P2010 ISR
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P20
10
0.5
1
1.5
13 / 26
P2010 Tune + ISR Parameters Variation - 2
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 ISR
P2010 tune
↑P2010 ISR
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P20
10
0.5
1
1.550 100 150 200 250
N (
Log
scal
e)
210
310
410
510
610 ↓P2010 ISR P2010 tune
↑P2010 ISR
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250
dist
./P20
10
0.5
1
1.5
Effect on high-pt jet #, pt ∼ as in the case of Psoft and Phard ISR parametersonly variation.
14 / 26
P2010 Tune + ISR Parameters Variation - 3
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
N
5000
10000
15000↓P2010 ISR
P2010 tune
↑P2010 ISR
ηleading jet
ηjet -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
dist
./P20
10
0.5
1
1.50.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
E(r
)/E
(R)
0.5
1
↓P2010 ISR
P2010 tune
↑P2010 ISR
leading jet E-profile
r (R=0.4)0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
dist
./P20
10
0.8
1
1.2
No larga η shape modulation observed in the central region.
Effect on jet energy flow distribution observed, but smaller than the effects due toFSR, UE.
15 / 26
P2010 Tune + FSR Parameters Variation - 1
Producing 2 additional samples with P2010 + (only) FSR λ (PARP(72)) variedon top:
P2010: PARP(72)=0.26,P2010 FSR ↑: PARP(72)=0.52,P2010 FSR ↓: PARP(72)=0.13.
0 1 2 3 4 5 6
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 FSR
P2010 tune
↑P2010 FSR
# of jets with pt>20 GeV
# of jets0 1 2 3 4 5 6
dist
./P20
10
0.5
1
1.50 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 FSR
P2010 tune
↑P2010 FSR
# of jets with pt>30 GeV
# of jets0 1 2 3 4
dist
./P20
10
0.5
1
1.5
16 / 26
P2010 Tune + FSR Parameters Variation - 2
Producing 2 additional samples with P2010 + (only) FSR λ (PARP(72)) variedon top:
P2010: PARP(72)=0.26,P2010 FSR ↑: PARP(72)=0.52,P2010 FSR ↓: PARP(72)=0.13.
0 1 2 3 4
N (
Log
scal
e)
210
310
410
510
610
710↓P2010 FSR
P2010 tune
↑P2010 FSR
# of jets with pt>40 GeV
# of jets0 1 2 3 4
dist
./P20
10
0.5
1
1.550 100 150 200 250
N (
Log
scal
e)
210
310
410
510
610 ↓P2010 FSR P2010 tune
↑P2010 FSR
(inclusive) jet pt
jet pt [GeV]50 100 150 200 250
dist
./P20
10
0.5
1
1.5
17 / 26
P2010 Tune + FSR Parameters Variation 1
Producing 2 additional samples with P2010 + (only) FSR λ (PARP(72)) variedon top:
P2010: PARP(72)=0.26,P2010 FSR ↑: PARP(72)=0.52,P2010 FSR ↓: PARP(72)=0.13.
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
E(r
)/E
(R)
0.5
1
↓P2010 FSR
P2010 tune
↑P2010 FSR
leading jet E-profile
r (R=0.4)0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
dist
./P20
10
0.8
1
1.20.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
E(r
)/E
(R)
0.5
1
↓P2010 FSR
P2010 tune
↑P2010 FSR
2nd hard. jet E-profile
r (R=0.4)0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
dist
./P20
10
0.8
1
1.2
Also in matched samples: more FSR ⇒ broader jets.
18 / 26
P2010 Tune + FSR Parameters Variation - 3
Producing 2 additional samples with P2010 + (only) FSR λ (PARP(72)) variedon top:
P2010: PARP(72)=0.26,P2010 FSR ↑: PARP(72)=0.52,P2010 FSR ↓: PARP(72)=0.13.
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
E(r
)/E
(R)
0.5
1
↓P2010 FSR
P2010 tune
↑P2010 FSR
[20-30 GeV] jet E-profile
r (R=0.4)0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
dist
./P20
10
0.8
1
1.20.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
E(r
)/E
(R)
0.5
1
↓P2010 FSR
P2010 tune
↑P2010 FSR
[40-60 GeV] jet E-profile
r (R=0.4)0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
dist
./P20
10
0.8
1
1.2
Also in matched samples: more FSR ⇒ broader jets.
19 / 26
P2010 Tune + ISR/FSR Parameters Effect on W pt
0 50 100 150 200 250
N (
Log
scal
e)
1
10
210
310
410
510
610 ↓P2010 ISR P2010 tune
↑P2010 ISR
(stat.3) W pt
W pt [GeV] 0 50 100 150 200 250
dist
./P20
10
0.5
1
1.50 50 100 150 200 250
N (
Log
scal
e)
1
10
210
310
410
510
610 ↓P2010 FSR P2010 tune
↑P2010 FSR
(stat.3) W pt
W pt [GeV] 0 50 100 150 200 250
dist
./P20
10
0.5
1
1.5
ISR: effect in the matched events reversed wrt. Pythia standalone.
FSR: no significant effect.
20 / 26
Effects on x-Section (Matching Efficiency)
due to ISR and FSR variation the x-section values change for ∼ 10-20% wrt. thecentral sample (P2010).
the P2010 ISR ↑, P2010 ISR↓, P2010 FSR↑, P2010 FSR↓ variations are asdefined in previous slides.
more hard ISR activity decreases the x-section and the matching efficiency in allsub-samples,
more FSR increases the x-section for 0j, but decreases it for other sub-samples.As a result the effect of FSR variations on the total x-section is small.
tune 0j 1j 2j 3j 4j total [pb]Phard 7287 ± 3.9 728 ± 2.6 141 ± 1.3 27 ± 0.2 6.6 ± 0.2 8190 ± 8P0 7556± 3.6 814 ± 2.7 166 ± 1.3 32 ± 0.3 7.8 ± 0.3 8576 ± 8Psoft 7804 ± 3.4 944 ± 2.8 207 ± 1.5 42 ± 0.3 10.1± 0.3 9007 ± 8P2010 7596 ± 5.8 774 ± 2.7 153 ± 1.3 29 ± 0.2 6.9 ± 0.2 8559 ± 10P2010 ISR ↑ 7226 ± 6.8 693 ± 2.6 132 ± 1.2 25 ± 0.2 6.1 ± 0.2 8082 ± 11P2010 ISR↓ 7846 ± 4.9 822 ± 2.7 166 ± 1.3 32 ± 0.3 7.3 ± 0.3 8874 ± 10P2010 FSR↑ 7639 ± 5.7 733 ± 2.6 136 ± 1.2 24 ± 0.2 5.5 ± 0.2 8538 ± 10P2010 FSR↓ 7546 ± 6.0 809 ± 2.7 164 ± 1.3 33 ± 0.3 8.2 ± 0.3 8560 ± 11
21 / 26
Events Failing the Matching in P2010 Tune - 1
Look at pt of the hardest ISR, FSR and MI branching in events that pass and inevents that do not pass the matching (using P2010 +2 partons for the plots).pt of the hardest branching corresponds to Pythia parameter VINT(357) (ISR)and VINT(358) (FSR), Pythia Manual has the exact definitions.
ISVETO=0 : EVENT PASSES, red histogramISVETO!=0 : EVENT FAILS 1
pt [GeV] 0 20 40 60 80 100 120 140 160
N/N
∆
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7 ISVETO = 0
ISVETO = 1
ISVETO = 2
ISVETO = 3
hardest ISR branching pt
pt [GeV] 0 20 40 60 80 100 120 140 160
N/N
∆
00.10.20.30.40.50.60.70.80.9
ISVETO = 0
ISVETO = 1
ISVETO = 2
ISVETO = 3
hardest FSR branching pt
We see that:hard ISR emission : event will more likely fail the matching,hard FSR emission : the event will more likely fail the matching.
1since:
ISVETO=1 : there are fewer reconstructed jets than partons,ISVETO=2 : not all partons match a jet,ISVETO=3 : there are additional jets in an exclusive sample.ISVETO=4 : the matched jets in inclusive sample are not the hardest ones.
22 / 26
Events Failing the Matching in P2010 Tune - 2
ISVETO=0 : EVENT PASSES
ISVETO!=0 : EVENT FAILS 2
pt of the hardest MI branching corresponds to Pythia parameter VINT(356),Pythia Manual has the exact definition.
pt [GeV] 0 20 40 60 80 100
N/N
(Lo
g S
cale
) ∆
-410
-310
-210
-110
ISVETO = 0
ISVETO = 1
ISVETO = 2
ISVETO = 3
hardest MI branching pt
We see that:whether event fails or passes the matching does not depend on the hard(est) MI.
2since:
ISVETO=1 : there are fewer reconstructed jets than partons,ISVETO=2 : not all partons match a jet,ISVETO=3 : there are additional jets in an exclusive sample.ISVETO=4 : the matched jets in inclusive sample are not the hardest ones.
23 / 26
Summary of the Observations
When Pythia parameters are varied for Alpgen + Pythia event generation:increasing hard ISR or hard FSR activity:
will affect the matching efficiency, x-section,will result in softer hard jet # distributions and pt-spectra, which is vice-versa as the effectsobtained with Pythia standalone,some trends will stay as observed in Pythia stand. (e.g. more FSR will still result in broader jets. . .).
UE parameters variations:have not been observed to have a significant effect on matching efficiency, x-section,will (not) push jets above the (relatively low, e.g. 20 GeV) pt threshold, which is also observedin Pythia standalone.
24 / 26
Thank you
25 / 26
What is MSTP(72), cf. Pythia Manual
26 / 26