Monte Carlo simulation for the LHC
Monte Carlo simulation for the LHC
Monte Carlo simulation for the LHC
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong><br />
Johan Alwall<br />
National Taiwan University and Fermilab<br />
Lecture 1<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall
•<br />
•<br />
Lecture I:<br />
Outline of lectures<br />
➡ New Physics at hadron colliders<br />
➡ Simulation of collider events<br />
➡ Parton Showers<br />
➡ Jet matching between ME and PS<br />
Lecture II:<br />
➡ The NLO revolution<br />
➡ NLO+Parton Showers<br />
➡ Workflow <strong>for</strong> BSM phenomenology<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
2
•<br />
Aims <strong>for</strong> <strong>the</strong>se lectures<br />
Get you acquainted with <strong>the</strong> concepts and tools<br />
used in event <strong>simulation</strong> at hadron colliders<br />
• Answer as many of your questions as I can<br />
(so please ask questions!)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
3
18<br />
I%6B12*B1$@<br />
Why <strong>the</strong> <strong>LHC</strong>?<br />
•<br />
) %6;< 9.*(27 ;H %>1?@<br />
) 6(*+,-*.@<br />
!"##$<br />
;#5&23)&4#56.76<<br />
•<br />
•<br />
) 6(11B1B@<br />
!"##$<br />
) %6;1?@<br />
Higgs boson mass “naturally” at<br />
mass of new physics<br />
! !"##$%&'$'(%)*$$%(*+,-*../%<br />
(only known “NP scale”: Planck scale at<br />
~1018 *+%+01%$2*.1%'3%(14%50/$"2$%<br />
GeV) 6'(./%7('4(8%9.*(27%$2*.1:%<br />
;< ;= %>1?@<br />
Standard Model only “works” if<br />
! A+*(B*-B%C'B1.%'(./%4'-7$%"3%<br />
Higgs mass below ~800 GeV<br />
+01%!"##$%)*$$%&1.'4%D=
18<br />
N#C1.$/1.-E<br />
, G0/&$F #C9: 9H #D.BE<br />
, DIA #C9: 9J #D.BE#K<br />
Why <strong>the</strong> <strong>LHC</strong>?<br />
;#5&23)&4#56.76<<br />
L.=#G82-)$-M<br />
, C&..1.1E<br />
")77-<br />
, #C9::#D.BE<br />
=./F<br />
, >(%'%& #C9#D.BE<br />
• ΔMH contribution must be<br />
!! #$%&'()*+')%&#,+-'#<br />
canceled " by bare mass term. For<br />
*.#$/&$.0.1#*2#*/(.#<br />
,/--#'.(,3#4%(#5)&.6<br />
'+&)&7#0.--#'8/	:;82-)$-#<br />
=8)$8#$+'-#%55#'8.#<br />
?+/1(/')$#0%%>-#/'#<br />
@9#A.B<br />
fine-tuning less than 1%, need<br />
new physics which cuts off <strong>the</strong><br />
quadratic loops at ~1 TeV<br />
@9: O #D.B P 9: J #D.B P<br />
!"#$%&'()$((&*&+,%-.%/&0"1&23)&4#56.76&$-&-#3&8+9 :<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
5
Why <strong>the</strong> <strong>LHC</strong>?<br />
The Hierarchy problem, toge<strong>the</strong>r with Dark Matter (and to<br />
some extent Grand Unification) have been driving New<br />
Physics model building in past 30 years<br />
➡ Supersymmetry<br />
➡ Large Extra Dimensions<br />
➡ Randall-Sundrum (warped extra dimensions)<br />
➡ Little Higgs <strong>the</strong>ories<br />
➡ Composite models/Technicolor/Topcolor/...<br />
➡ ... (mostly variants/combinations)<br />
But of course, we might also find something<br />
completely unexpected!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
6
•<br />
•<br />
•<br />
New Physics at hadron colliders<br />
The <strong>LHC</strong> has taken over from <strong>the</strong> Tevatron!<br />
Significant luminocities<br />
➡ Tevatron collected >10 fb -1 in <strong>the</strong> last 10 years<br />
➡ Fantastic legacy, including several interesting<br />
excesses!<br />
➡ <strong>LHC</strong> already has a spectacular 10 fb -1 !<br />
(perhaps as much as 35 fb -1 by end of this run!)<br />
➡ Allows ever-more stringent tests of <strong>the</strong> SM!<br />
➡ Already found what might be <strong>the</strong> Higgs boson!<br />
How interpret excesses? How determine Standard<br />
Model backgrounds?<br />
➡ <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong>! (combined with data-driven methods)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7
Example: CDF excess in W + 2 jets<br />
WW, WZ<br />
W + Nobody<br />
knows what?<br />
)<br />
2<br />
Events/(8 GeV/c<br />
150<br />
100<br />
50<br />
0<br />
-50<br />
CDF collaboration, arXiv:1104.0699<br />
Bkg Sub Data (4.3 fb<br />
Gaussian<br />
WW+WZ<br />
100 200<br />
Mjj<br />
2<br />
[GeV/c ]<br />
Background subtracted data (except WW/WZ)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
(a)<br />
-1<br />
)<br />
8
Example: CDF excess in W + 2 jets<br />
A more complete picture<br />
)<br />
2<br />
Events/(8 GeV/c<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
CDF collaboration, arXiv:1104.0699<br />
100 200<br />
CDF data (4.3 fb<br />
Gaussian 2.5%<br />
WW+WZ 4.8%<br />
W+Jets 78.0%<br />
Top 6.3%<br />
Z+jets 2.8%<br />
QCD 5.1%<br />
[GeV/c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
M<br />
jj<br />
-1<br />
)<br />
(c)<br />
2<br />
]<br />
9
Example: CDF excess in W + 2 jets<br />
CDF data<br />
A more complete picture<br />
)<br />
2<br />
Events/(8 GeV/c<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
CDF collaboration, arXiv:1104.0699<br />
100 200<br />
CDF data (4.3 fb<br />
Gaussian 2.5%<br />
WW+WZ 4.8%<br />
W+Jets 78.0%<br />
Top 6.3%<br />
Z+jets 2.8%<br />
QCD 5.1%<br />
[GeV/c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
M<br />
jj<br />
-1<br />
)<br />
(c)<br />
2<br />
]<br />
9
Example: CDF excess in W + 2 jets<br />
CDF data<br />
Standard Model<br />
backgrounds<br />
(shape from <strong>simulation</strong>)<br />
A more complete picture<br />
)<br />
2<br />
Events/(8 GeV/c<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
CDF collaboration, arXiv:1104.0699<br />
100 200<br />
CDF data (4.3 fb<br />
Gaussian 2.5%<br />
WW+WZ 4.8%<br />
W+Jets 78.0%<br />
Top 6.3%<br />
Z+jets 2.8%<br />
QCD 5.1%<br />
[GeV/c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
M<br />
jj<br />
-1<br />
)<br />
(c)<br />
2<br />
]<br />
9
Example: CDF excess in W + 2 jets<br />
CDF data<br />
Excess<br />
Standard Model<br />
backgrounds<br />
(shape from <strong>simulation</strong>)<br />
A more complete picture<br />
)<br />
2<br />
Events/(8 GeV/c<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
CDF collaboration, arXiv:1104.0699<br />
100 200<br />
CDF data (4.3 fb<br />
Gaussian 2.5%<br />
WW+WZ 4.8%<br />
W+Jets 78.0%<br />
Top 6.3%<br />
Z+jets 2.8%<br />
QCD 5.1%<br />
[GeV/c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
M<br />
jj<br />
-1<br />
)<br />
(c)<br />
2<br />
]<br />
9
Example: CDF excess in W + 2 jets<br />
CDF data<br />
Excess<br />
Standard Model<br />
backgrounds<br />
(shape from <strong>simulation</strong>)<br />
A more complete picture<br />
)<br />
2<br />
Events/(8 GeV/c<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
100 200<br />
We certainly need to know our backgrounds well!<br />
CDF collaboration, arXiv:1104.0699<br />
CDF data (4.3 fb<br />
Gaussian 2.5%<br />
WW+WZ 4.8%<br />
W+Jets 78.0%<br />
Top 6.3%<br />
Z+jets 2.8%<br />
QCD 5.1%<br />
[GeV/c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
M<br />
jj<br />
-1<br />
)<br />
(c)<br />
2<br />
]<br />
9
Processes at Hadron Colliders<br />
First: Understand our processes!<br />
Cross sections at a collider depend on:<br />
• Coupling strength<br />
• Coupling to what?<br />
(light quarks, gluons, heavy quarks,<br />
EW gauge bosons?)<br />
• Mass<br />
• Single production/pair production<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
b<br />
W<br />
Z<br />
t<br />
10
Processes at Hadron Colliders<br />
First: Understand our processes!<br />
Cross sections at a collider depend on:<br />
• Coupling strength<br />
• Coupling to what?<br />
(light quarks, gluons, heavy quarks,<br />
EW gauge bosons?)<br />
• Mass<br />
• Single production/pair production<br />
Expected<br />
new<br />
physics<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
b<br />
W<br />
Z<br />
t<br />
10
Master <strong>for</strong>mula<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
11
•<br />
ˆσab→X(ˆs, . . .)<br />
Parton level<br />
cross section<br />
Master <strong>for</strong>mula<br />
Parton level cross section from matrix element<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
11
•<br />
•<br />
Master <strong>for</strong>mula<br />
ˆσab→X(ˆs, . . .) fa(x1)fb(x2)<br />
Parton level<br />
cross section<br />
Parton density<br />
functions<br />
Parton level cross section from matrix element<br />
Parton density (or distribution) functions:<br />
Process independent, determined by particle type<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
11
•<br />
•<br />
Tevatron vs. <strong>the</strong> <strong>LHC</strong><br />
Tevatron: 2 TeV proton-antiproton collider<br />
⎯⎯ ⎯⎯<br />
➡ Most important: q-q annihilation (85% of t t )<br />
<strong>LHC</strong>: 8-14 TeV proton-proton collider<br />
➡ Most important: g-g annihilation (90% of t t )<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
⎯⎯<br />
12
•<br />
•<br />
Tevatron vs. <strong>the</strong> <strong>LHC</strong><br />
Tevatron: 2 TeV proton-antiproton collider<br />
⎯⎯ ⎯⎯<br />
➡ Most important: q-q annihilation (85% of t t )<br />
<strong>LHC</strong>: 8-14 TeV proton-proton collider<br />
➡ Most important: g-g annihilation (90% of t t )<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
⎯⎯<br />
12
inematics<br />
atically towards low x<br />
ions <strong>for</strong> <strong>LHC</strong>, e.g.<br />
Parton densities<br />
Ratio of Luminosity: <strong>LHC</strong> at 7 TeV vs Tevatron<br />
pdf’s measured in deep-inelastic scattering!<br />
!! Power of collider can be<br />
fully characterized by ratio<br />
of parton luminosities<br />
!! Ratio larger <strong>for</strong> gg than qq<br />
!! Due to steap rise of gluon<br />
towards low x<br />
!! M X =100 GeV<br />
!! gg: R!10, e.g. Higgs<br />
!! qq: R!3, e.g. W and Z<br />
!! M X=800 GeV<br />
!! gg: R!1000, e.g. SUSY<br />
At small x (small ŝ), gluon domination.<br />
At large x valence quarks<br />
!! qq: R!20, e.g. Z’<br />
<strong>LHC</strong> <strong>for</strong>midable at large mass –<br />
For low mass, Tevatron backgrounds smaller<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 11!<br />
13
Back to <strong>the</strong> processes<br />
<strong>LHC</strong> at 7 TeV vs Tevatron<br />
e<br />
atio<br />
qq<br />
n<br />
Z<br />
SY<br />
PreSUSY, Beijing, August 8-11, 2012 11! <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 14
Back to <strong>the</strong> processes<br />
<strong>LHC</strong> at 7 TeV vs Tevatron<br />
e<br />
atio<br />
qq<br />
n<br />
Z<br />
SY<br />
PreSUSY, Beijing, August 8-11, 2012 11! <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 14
Back to <strong>the</strong> processes<br />
<strong>LHC</strong> at 7 TeV vs Tevatron<br />
e<br />
atio<br />
qq<br />
n<br />
Z<br />
SY<br />
PreSUSY, Beijing, August 8-11, 2012 11! <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 14
Back to <strong>the</strong> processes<br />
<strong>LHC</strong> at 7 TeV vs Tevatron<br />
e<br />
atio<br />
qq<br />
n<br />
Z<br />
SY<br />
PreSUSY, Beijing, August 8-11, 2012 11! <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 14
Simulation of collider events<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
15
Sherpa artist<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
16
2<br />
1. High-Q Scattering<br />
2. Parton Shower<br />
3. Hadronization 4. Underlying Event<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
17
2<br />
1. High-Q Scattering<br />
2. Parton Shower<br />
3. Hadronization 4. Underlying Event<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
18
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall 21
2<br />
1. High-Q Scattering<br />
2. Parton Shower<br />
3. Hadronization 4. Underlying Event<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
22
List of processes<br />
implemented<br />
in Pythia (by hand!)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
23
Automated Matrix Element Generators<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
24
•<br />
Automated Matrix Element Generators<br />
High-Q2 scattering processes: In principle infinite number<br />
of processes <strong>for</strong> innumerable number of models<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
24
•<br />
•<br />
Automated Matrix Element Generators<br />
High-Q2 scattering processes: In principle infinite number<br />
of processes <strong>for</strong> innumerable number of models<br />
Implementation by hand time-consuming, labor intensive<br />
and error prone<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
24
•<br />
•<br />
•<br />
Automated Matrix Element Generators<br />
High-Q2 scattering processes: In principle infinite number<br />
of processes <strong>for</strong> innumerable number of models<br />
Implementation by hand time-consuming, labor intensive<br />
and error prone<br />
Instead: Automated matrix element generators<br />
➡ Use Feynman rules to build diagrams<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
24
•<br />
•<br />
•<br />
•<br />
Automated Matrix Element Generators<br />
High-Q2 scattering processes: In principle infinite number<br />
of processes <strong>for</strong> innumerable number of models<br />
Implementation by hand time-consuming, labor intensive<br />
and error prone<br />
Instead: Automated matrix element generators<br />
➡ Use Feynman rules to build diagrams<br />
Given files defining <strong>the</strong> model content: particles,<br />
parameters and interactions, allows to generate any<br />
process <strong>for</strong> a given model!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
24
Automated Matrix Element Generators<br />
•<br />
•<br />
Automatic matrix element generators:<br />
➡ CalcHep / CompHep<br />
➡ MadGraph<br />
➡ AMEGIC++ (Sherpa)<br />
➡ Whizard<br />
Standard Model only, with fast matrix elements <strong>for</strong> high<br />
parton multiplicity final states:<br />
➡ AlpGen<br />
➡ HELAC<br />
➡ COMIX (Sherpa)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
25
2<br />
1. High-Q Scattering<br />
2. Parton Shower<br />
3. Hadronization 4. Underlying Event<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
26
Parton Shower MC event generators<br />
• General-purpose tools<br />
• Complete exclusive description of <strong>the</strong> events: hard scattering,<br />
showering, hadronization, underlying event<br />
• Reliable and well tuned to experimental data.<br />
most well-known: PYTHIA, HERWIG, SHERPA<br />
• Significant progress in <strong>the</strong> development of new showering algorithms<br />
with <strong>the</strong> final aim to go to NLO in QCD<br />
[Nagy, Soper, 2005; Giele, Kosower, Skands, 2007; Krauss, Schumann, 2007]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
27
Parton Shower basics<br />
The spin averaged (unregulated) splitting functions <strong>for</strong> <strong>the</strong> various<br />
types of branching are:<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
29
Parton Shower basics<br />
The spin averaged (unregulated) splitting functions <strong>for</strong> <strong>the</strong> various<br />
types of branching are:<br />
Comments:<br />
* Gluons radiate <strong>the</strong> most<br />
* There are soft divergences in z=1 and z=0.<br />
* Pqg has no soft divergences.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
29
Final-state parton showers<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
1. Start <strong>the</strong> evolution at <strong>the</strong> virtual mass scale t0 (e.g. <strong>the</strong> mass of <strong>the</strong><br />
decaying particle) and momentum fraction z0 = 1<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
1. Start <strong>the</strong> evolution at <strong>the</strong> virtual mass scale t0 (e.g. <strong>the</strong> mass of <strong>the</strong><br />
decaying particle) and momentum fraction z0 = 1<br />
2. Given a virtual mass scale ti and momentum fraction xi at some<br />
stage in <strong>the</strong> evolution, generate <strong>the</strong> scale of <strong>the</strong> next emission ti+1<br />
according to <strong>the</strong> Sudakov probability ∆(ti,ti+1) by solving<br />
∆(ti+1,ti) = R<br />
where R is a random number (uni<strong>for</strong>m on [0, 1]).<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
1. Start <strong>the</strong> evolution at <strong>the</strong> virtual mass scale t0 (e.g. <strong>the</strong> mass of <strong>the</strong><br />
decaying particle) and momentum fraction z0 = 1<br />
2. Given a virtual mass scale ti and momentum fraction xi at some<br />
stage in <strong>the</strong> evolution, generate <strong>the</strong> scale of <strong>the</strong> next emission ti+1<br />
according to <strong>the</strong> Sudakov probability ∆(ti,ti+1) by solving<br />
∆(ti+1,ti) = R<br />
where R is a random number (uni<strong>for</strong>m on [0, 1]).<br />
3. If ti+1 < tcut it means that <strong>the</strong> shower has finished.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
1. Start <strong>the</strong> evolution at <strong>the</strong> virtual mass scale t0 (e.g. <strong>the</strong> mass of <strong>the</strong><br />
decaying particle) and momentum fraction z0 = 1<br />
2. Given a virtual mass scale ti and momentum fraction xi at some<br />
stage in <strong>the</strong> evolution, generate <strong>the</strong> scale of <strong>the</strong> next emission ti+1<br />
according to <strong>the</strong> Sudakov probability ∆(ti,ti+1) by solving<br />
∆(ti+1,ti) = R<br />
where R is a random number (uni<strong>for</strong>m on [0, 1]).<br />
3. If ti+1 < tcut it means that <strong>the</strong> shower has finished.<br />
4. O<strong>the</strong>rwise, generate z = zi/zi+1 with a distribution proportional to<br />
(αs/2π)P(z), where P(z) is <strong>the</strong> appropriate splitting function.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
With <strong>the</strong> Sudakov <strong>for</strong>m factor, we can now implement a finalstate<br />
parton shower in a <strong>Monte</strong> <strong>Carlo</strong> event generator!<br />
1. Start <strong>the</strong> evolution at <strong>the</strong> virtual mass scale t0 (e.g. <strong>the</strong> mass of <strong>the</strong><br />
decaying particle) and momentum fraction z0 = 1<br />
2. Given a virtual mass scale ti and momentum fraction xi at some<br />
stage in <strong>the</strong> evolution, generate <strong>the</strong> scale of <strong>the</strong> next emission ti+1<br />
according to <strong>the</strong> Sudakov probability ∆(ti,ti+1) by solving<br />
∆(ti+1,ti) = R<br />
where R is a random number (uni<strong>for</strong>m on [0, 1]).<br />
3. If ti+1 < tcut it means that <strong>the</strong> shower has finished.<br />
4. O<strong>the</strong>rwise, generate z = zi/zi+1 with a distribution proportional to<br />
(αs/2π)P(z), where P(z) is <strong>the</strong> appropriate splitting function.<br />
5. For each emitted particle, iterate steps 2-4 until branching stops.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
31
Final-state parton showers<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
32
Final-state parton showers<br />
• The result is a “cascade” or “shower” of partons with ever<br />
smaller virtualities.<br />
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 1: QCD<br />
Johan Alwall<br />
Plan of <strong>the</strong> lectures<br />
Introduction: The<br />
big picture<br />
Infrared Behaviour<br />
of QCD<br />
Jet Definitions<br />
Parton Showers<br />
Parton branchings<br />
Evolution<br />
equations and<br />
parton densities<br />
Logarithmic<br />
resummation<br />
Sudakov <strong>for</strong>m<br />
factors<br />
Angular ordering<br />
NLL Sudakovs<br />
Parton showers in<br />
<strong>Monte</strong> <strong>Carlo</strong>s<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
e -<br />
e +<br />
t0<br />
Due to <strong>the</strong>se successive branchings, <strong>the</strong> parton cascade or parton show<br />
develops. Each outgoing line is a source of a new cascade, until all ou<br />
lines have stopped branching. At this stage, which depends on <strong>the</strong> cut<br />
outgoing partons have to be converted into hadrons via a hadronizatio<br />
32
Final-state parton showers<br />
• The result is a “cascade” or “shower” of partons with ever<br />
smaller virtualities.<br />
• The cutoff scale tcut is usually set close to 1 GeV,<br />
<strong>the</strong> scale where non-perturbative effects start dominating<br />
Matching of Matrix<br />
over <strong>the</strong> perturbative Elements parton and shower.<br />
Parton Showers<br />
Lecture 1: QCD<br />
Johan Alwall<br />
Plan of <strong>the</strong> lectures<br />
Introduction: The<br />
big picture<br />
Infrared Behaviour<br />
of QCD<br />
Jet Definitions<br />
Parton Showers<br />
Parton branchings<br />
Evolution<br />
equations and<br />
parton densities<br />
Logarithmic<br />
resummation<br />
Sudakov <strong>for</strong>m<br />
factors<br />
Angular ordering<br />
NLL Sudakovs<br />
Parton showers in<br />
<strong>Monte</strong> <strong>Carlo</strong>s<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
e -<br />
e +<br />
t0<br />
tcut<br />
Due to <strong>the</strong>se successive branchings, <strong>the</strong> parton cascade or parton show<br />
develops. Each outgoing line is a source of a new cascade, until all ou<br />
lines have stopped branching. At this stage, which depends on <strong>the</strong> cut<br />
outgoing partons have to be converted into hadrons via a hadronizatio<br />
32
Final-state parton showers<br />
• The result is a “cascade” or “shower” of partons with ever<br />
smaller virtualities.<br />
• The cutoff scale tcut is usually set close to 1 GeV,<br />
<strong>the</strong> scale where non-perturbative effects start dominating<br />
over <strong>the</strong> perturbative parton shower.<br />
• At this point, phenomenological<br />
models are used to simulate<br />
how <strong>the</strong> partons turn into<br />
color-neutral hadrons.<br />
Hadronization not sensitive to<br />
<strong>the</strong> physics at scale t0, but only tcut!<br />
(can be tuned once and <strong>for</strong> all!)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
e -<br />
e +<br />
t0<br />
32
•<br />
•<br />
•<br />
•<br />
Initial-state parton splittings<br />
So far, we have looked at final-state (time-like)<br />
splittings<br />
For initial state, <strong>the</strong> splitting functions are <strong>the</strong> same<br />
However, <strong>the</strong>re is ano<strong>the</strong>r ingredient - <strong>the</strong> parton<br />
density (or distribution) functions (PDFs)<br />
➡ Probability to find a given parton in a hadron at a<br />
given momentum fraction x = pz/Pz and scale t<br />
How do <strong>the</strong> PDFs evolve with increasing t?<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
33
•<br />
•<br />
•<br />
Parton Shower MC event generators<br />
In both initial-state and final-state showers, <strong>the</strong> definition of t is<br />
not unique, as long as it has <strong>the</strong> dimension of scale:<br />
Different parton shower generators have made different choices:<br />
➡ Ariadne: “dipole pT”<br />
➡ Herwig: E⋅θ<br />
➡ Pythia (old): virtuality q 2<br />
➡ Pythia 6.4 and Pythia 8: pT<br />
➡ Sherpa: v. 1.1.x virtuality q 2 , v. 1.2.x “dipole pT”<br />
Note that all of <strong>the</strong> above are complete MC event generators<br />
with matrix elements, parton showers, hadronization, decay, and<br />
underlying event <strong>simulation</strong>.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
37
From Parton Showers to Hadronization<br />
• The parton shower evolves <strong>the</strong> hard scattering down to <strong>the</strong> scale of<br />
O(1GeV).<br />
• At this scale, QCD is no longer perturbative. some hadronization model is<br />
used to describe <strong>the</strong> transition from <strong>the</strong> perturbative PS region to <strong>the</strong><br />
non-perturbative hadronization region.<br />
• Main hadronization models:<br />
➡ String hadronization (Pythia)<br />
➡ Cluster hadronization (Herwig)<br />
• Hadronization only acts locally, not sensitive to high-q 2 scattering.<br />
e -<br />
e +<br />
[Andersson,Gustafson,Ingelman,Sjöstrand (1983)]<br />
[Webber (1984)]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
38
•<br />
•<br />
•<br />
Detector <strong>simulation</strong><br />
Detector <strong>simulation</strong><br />
➡ Fast general-purpose detector simulators:<br />
Delphes, PGS (“Pretty good <strong>simulation</strong>s”), AcerDet<br />
➡ Specify parameters to simulate different experiments<br />
Experiment-specific fast <strong>simulation</strong><br />
➡ Detector response parameterized<br />
➡ Run time ms-s/event<br />
Experiment-specific full <strong>simulation</strong><br />
➡ Full tracking of particles through detector using GEANT<br />
➡ Run time several minutes/event<br />
➡ See Michael’s lecture <strong>for</strong> much more details!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
39
Back to our favorite piece of art!<br />
2 1. High-Q Scattering<br />
2. Parton Shower<br />
How do we define <strong>the</strong> limit between parton shower<br />
and matrix element?<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
40
Matrix Elements vs. Parton Showers<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
41
Matrix Elements vs. Parton Showers<br />
ME<br />
1. Fixed order calculation<br />
2. Computationally expensive<br />
3. Limited number of particles<br />
4. Valid when partons are hard and<br />
well separated<br />
5. Quantum interference correct<br />
6. Needed <strong>for</strong> multi-jet description<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
41
Matrix Elements vs. Parton Showers<br />
ME<br />
1. Fixed order calculation<br />
2. Computationally expensive<br />
3. Limited number of particles<br />
4. Valid when partons are hard and<br />
well separated<br />
5. Quantum interference correct<br />
6. Needed <strong>for</strong> multi-jet description<br />
Shower MC<br />
1. Resums logs to all orders<br />
2. Computationally cheap<br />
3. No limit on particle multiplicity<br />
4. Valid when partons are collinear<br />
and/or soft<br />
5. Partial interference through<br />
angular ordering<br />
6. Needed <strong>for</strong> hadronization<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
41
Matrix Elements vs. Parton Showers<br />
ME<br />
1. Fixed order calculation<br />
2. Computationally expensive<br />
3. Limited number of particles<br />
4. Valid when partons are hard and<br />
well separated<br />
5. Quantum interference correct<br />
6. Needed <strong>for</strong> multi-jet description<br />
Shower MC<br />
1. Resums logs to all orders<br />
2. Computationally cheap<br />
3. No limit on particle multiplicity<br />
4. Valid when partons are collinear<br />
and/or soft<br />
5. Partial interference through<br />
angular ordering<br />
6. Needed <strong>for</strong> hadronization<br />
Approaches are complementary: merge <strong>the</strong>m!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
41
Matrix Elements vs. Parton Showers<br />
ME<br />
1. Fixed order calculation<br />
2. Computationally expensive<br />
3. Limited number of particles<br />
4. Valid when partons are hard and<br />
well separated<br />
5. Quantum interference correct<br />
6. Needed <strong>for</strong> multi-jet description<br />
Shower MC<br />
1. Resums logs to all orders<br />
2. Computationally cheap<br />
3. No limit on particle multiplicity<br />
4. Valid when partons are collinear<br />
and/or soft<br />
5. Partial interference through<br />
angular ordering<br />
6. Needed <strong>for</strong> hadronization<br />
Approaches are complementary: merge <strong>the</strong>m!<br />
Difficulty: avoid double counting, ensure smooth distributions<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
41
(pb/bin)<br />
T<br />
/dP<br />
!<br />
d<br />
10<br />
10<br />
10<br />
10<br />
1<br />
-1<br />
-2<br />
-3<br />
PS alone vs ME matching<br />
In a matched sample <strong>the</strong>se differences are irrelevant since <strong>the</strong> behavior<br />
at high pt is dominated by <strong>the</strong> matrix element.<br />
Q<br />
Q<br />
P<br />
P<br />
2<br />
2<br />
2<br />
T<br />
2<br />
T<br />
(wimpy)<br />
(power)<br />
(wimpy)<br />
(power)<br />
tt+0,1,2,3<br />
partons + Pythia (MMLM)<br />
of <strong>the</strong> 2-nd extra jet<br />
0 50 100 150 200 250 300 350 400<br />
GeV<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
P<br />
T<br />
[MadGraph]<br />
43
Goal <strong>for</strong> ME-PS merging/matching<br />
Matrix element<br />
Parton shower<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
44
•<br />
Goal <strong>for</strong> ME-PS merging/matching<br />
Regularization of matrix element divergence<br />
Matrix element<br />
Parton shower<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
44
•<br />
•<br />
Goal <strong>for</strong> ME-PS merging/matching<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Matrix element<br />
Parton shower<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
44
•<br />
•<br />
•<br />
Goal <strong>for</strong> ME-PS merging/matching<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
44
•<br />
•<br />
•<br />
Goal <strong>for</strong> ME-PS merging/matching<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Desired curve<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
44
ME<br />
↓<br />
Merging ME with PS<br />
PS →<br />
...<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
...<br />
[Mangano]<br />
[Catani, Krauss, Kuhn, Webber]<br />
[Lönnblad]<br />
45
ME<br />
↓<br />
Merging ME with PS<br />
PS →<br />
DC DC<br />
DC<br />
...<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
...<br />
[Mangano]<br />
[Catani, Krauss, Kuhn, Webber]<br />
[Lönnblad]<br />
45
ME<br />
↓<br />
Merging ME with PS<br />
kT > Q c<br />
kT > Q c<br />
kT > Q c<br />
PS →<br />
...<br />
kT < Q c<br />
kT < Q c<br />
kT > Q c<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
...<br />
[Mangano]<br />
[Catani, Krauss, Kuhn, Webber]<br />
[Lönnblad]<br />
kT < Q c<br />
kT < Q c<br />
45
ME<br />
↓<br />
Merging ME with PS<br />
kT > Q c<br />
kT > Q c<br />
kT > Q c<br />
PS →<br />
...<br />
kT < Q c<br />
kT < Q c<br />
kT > Q c<br />
Double counting between ME and PS easily avoided using phase space<br />
cut between <strong>the</strong> two: PS below cutoff, ME above cutoff.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
...<br />
[Mangano]<br />
[Catani, Krauss, Kuhn, Webber]<br />
[Lönnblad]<br />
kT < Q c<br />
kT < Q c<br />
45
•<br />
•<br />
•<br />
Merging ME with PS<br />
So double counting problem easily solved, but<br />
what about getting smooth distributions that are<br />
independent of <strong>the</strong> precise value of Q c ?<br />
Below cutoff, distribution is given by PS<br />
- need to make ME look like PS near cutoff<br />
Let’s take ano<strong>the</strong>r look at <strong>the</strong> PS!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
46
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
t0<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
tcut tcut<br />
t2<br />
tcut<br />
tcut<br />
10 / 29<br />
47
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
• Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
How does <strong>the</strong> PS generate <strong>the</strong> configuration above?<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
t0<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
tcut tcut<br />
t2<br />
tcut<br />
tcut<br />
10 / 29<br />
47
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
• Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
How does <strong>the</strong> PS generate <strong>the</strong> configuration above?<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
• Probability <strong>for</strong> <strong>the</strong> splitting at t1 is given by<br />
new particle (e.g. q¯q → g, qg → q)<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
t0<br />
(∆q(t1,t0))<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
2 αs(t1)<br />
tcut tcut<br />
t2<br />
tcut<br />
tcut<br />
2π Pgq(z)<br />
10 / 29<br />
47
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
• Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
How does <strong>the</strong> PS generate <strong>the</strong> configuration above?<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
• Probability <strong>for</strong> <strong>the</strong> splitting at t1 is given by<br />
new particle (e.g. q¯q → g, qg → q)<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
t0<br />
(∆q(t1,t0))<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
2 αs(t1)<br />
tcut tcut<br />
t2<br />
tcut<br />
tcut<br />
2π Pgq(z)<br />
10 / 29<br />
47
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
• Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
How does <strong>the</strong> PS generate <strong>the</strong> configuration above?<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
• Probability <strong>for</strong> <strong>the</strong> splitting at t1 is given by<br />
new particle (e.g. q¯q → g, qg → q)<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
t0<br />
(∆q(t1,t0))<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
2 αs(t1)<br />
tcut tcut<br />
t2<br />
tcut<br />
tcut<br />
2π Pgq(z)<br />
10 / 29<br />
47
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
|M| 2 (ˆs, p3,p4,...)<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
10 / 29<br />
49
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
•<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
To get an equivalent treatment of <strong>the</strong> corresponding<br />
matrix element, do as follows:<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e − → q¯q)<br />
|M| 2 (ˆs, p3,p4,...)<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
10 / 29<br />
49
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
•<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
To get an equivalent treatment of <strong>the</strong> corresponding<br />
matrix element, do as follows:<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
1. Cluster <strong>the</strong> event using some clustering algorithm<br />
|M| 2 (ˆs, p3,p4,...)<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e− - this gives us a corresponding “parton → q¯q) shower history”<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
10 / 29<br />
49
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
•<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
To get an equivalent treatment of <strong>the</strong> corresponding<br />
matrix element, do as follows:<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
1. Cluster <strong>the</strong> event using some clustering algorithm<br />
t0<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e− - this gives us a corresponding “parton → q¯q) shower history”<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> branching.<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
t2 |M| 2 (ˆs, p3,p4,...)<br />
10 / 29<br />
49
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
•<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
To get an equivalent treatment of <strong>the</strong> corresponding<br />
matrix element, do as follows:<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
1. Cluster <strong>the</strong> event using some clustering algorithm<br />
t0<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e− - this gives us a corresponding “parton → q¯q) shower history”<br />
2. Reweight αs in each clustering vertex with <strong>the</strong> clustering<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT scale<br />
of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> 2 αs(t1) αs(t2)<br />
branching.<br />
|M| 2 → |M|<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
αs(t0)<br />
t2<br />
αs(t0)<br />
|M| 2 (ˆs, p3,p4,...)<br />
10 / 29<br />
49
Matching of Matrix<br />
Elements and<br />
Parton Showers<br />
Lecture 2:<br />
Matching in e + e −<br />
collisions<br />
Johan Alwall<br />
Why Matching?<br />
Present matching<br />
approaches<br />
CKKW matching in<br />
e + e − collisions<br />
Overview of <strong>the</strong><br />
CKKW procedure<br />
•<br />
Clustering <strong>the</strong><br />
n-jet event<br />
Sudakov<br />
reweighting<br />
Vetoed parton<br />
showers<br />
Highest<br />
multiplicity<br />
treatment<br />
Results of CKKW<br />
matching (Sherpa)<br />
Difficulties with<br />
practical<br />
implementations<br />
The MLM<br />
procedure<br />
Clustering <strong>the</strong> n-jet event<br />
Merging ME with PS<br />
To get an equivalent treatment of <strong>the</strong> corresponding<br />
matrix element, do as follows:<br />
1 Find <strong>the</strong> two partons with smallest jet separation yij<br />
1. Cluster <strong>the</strong> event using some clustering algorithm<br />
t0<br />
2 If partons allowed to cluster by QCD splitting rules: combine partons to<br />
new particle (e.g. q¯q → g, qg → q)<br />
3 Iterate 1-2 until 2 → 2 process reached (e + e− - this gives us a corresponding “parton → q¯q) shower history”<br />
2. Reweight αs in each clustering vertex with <strong>the</strong> clustering<br />
With <strong>the</strong> choice of <strong>the</strong> Durham jet measure, <strong>the</strong> jet separations di = √ yi Q0 at<br />
each branching corresponds closely to <strong>the</strong> kT scale<br />
of that branching, and is <strong>the</strong>re<strong>for</strong>e<br />
suitable to use as argument <strong>for</strong> αs in <strong>the</strong> 2 αs(t1) αs(t2)<br />
branching.<br />
3. Use some algorithm to apply <strong>the</strong> equivalent Sudakov<br />
suppression (∆q(tcut,t0)) 2 ∆g(t2,t1)(∆q(cut,t2)) 2<br />
|M| 2 → |M|<br />
αs(t0) αs(t0)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
t1<br />
t2<br />
|M| 2 (ˆs, p3,p4,...)<br />
10 / 29<br />
49
le of <strong>the</strong> procedure<br />
o simulate pp → W + jets.<br />
Matching <strong>for</strong> initial state radiation<br />
ck (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ck momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
x1<br />
x2<br />
PreSUSY, 2 Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
∆g (d2, dini) 2 αs(d2) αs(d1)<br />
50
Matching <strong>for</strong> initial state radiation<br />
• We are of course not interested in e+ e - but p-p(bar)<br />
le of <strong>the</strong> procedure<br />
o simulate pp → W + jets.<br />
- what happens <strong>for</strong> initial state radiation?<br />
ck (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ck momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
x1<br />
x2<br />
PreSUSY, 2 Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
∆g (d2, dini) 2 αs(d2) αs(d1)<br />
50
le of <strong>the</strong> procedure<br />
o simulate pp → W + jets.<br />
Matching <strong>for</strong> initial state radiation<br />
ck (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ck momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
x1<br />
x2<br />
PreSUSY, 2 Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
∆g (d2, dini) 2 αs(d2) αs(d1)<br />
52
Matching <strong>for</strong> initial state radiation<br />
• le of <strong>the</strong> procedure<br />
Again, use a clustering scheme to get a parton shower<br />
history<br />
o simulate pp → W + jets.<br />
ck (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ck momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
x1<br />
x2<br />
PreSUSY, 2 Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
∆g (d2, dini) 2 αs(d2) αs(d1)<br />
52
Matching <strong>for</strong> initial state radiation<br />
• le of <strong>the</strong> procedure<br />
Again, use a clustering scheme to get a parton shower<br />
history<br />
o simulate pp → W + jets.<br />
ck (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ck momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
x1<br />
t1 t2<br />
x2<br />
x1’<br />
t0<br />
PreSUSY, 2 Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
∆g (d2, dini) 2 αs(d2) αs(d1)<br />
52
•<br />
•<br />
Matching schemes<br />
We still haven’t specified how to apply <strong>the</strong> Sudakov<br />
reweighting to <strong>the</strong> matrix element<br />
Three general schemes available in <strong>the</strong> literature:<br />
➡ CKKW scheme [Catani,Krauss,Kuhn,Webber 2001; Krauss 2002]<br />
➡ Lönnblad scheme (or CKKW-L) [Lönnblad 2002]<br />
➡ MLM scheme [Mangano unpublished 2002; Mangano et al. 2007]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
53
CKKW matching<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
54
CKKW matching<br />
• Apply <strong>the</strong> required Sudakov suppression<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
analytically, using <strong>the</strong> best available (NLL) Sudakovs.<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
54
le of <strong>the</strong> procedure<br />
• Apply <strong>the</strong> required Sudakov suppression<br />
simulate pp → W + jets.<br />
k (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
k momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
CKKW matching<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
analytically, using <strong>the</strong> best available (NLL) Sudakovs.<br />
• Per<strong>for</strong>m “truncated showering”: Run <strong>the</strong> parton shower<br />
starting at t0, but <strong>for</strong>bid any showers above <strong>the</strong> cutoff scale tcut.<br />
t0<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
, dini))<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2) αs(d1)<br />
dini))<br />
αs(dini) αs(dini)<br />
54
le of <strong>the</strong> procedure<br />
• Apply <strong>the</strong> required Sudakov suppression<br />
simulate pp → W + jets.<br />
k (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
k momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
CKKW matching<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
analytically, using <strong>the</strong> best available (NLL) Sudakovs.<br />
• Per<strong>for</strong>m “truncated showering”: Run <strong>the</strong> parton shower<br />
starting at t0, but <strong>for</strong>bid any showers above <strong>the</strong> cutoff scale tcut.<br />
kT1<br />
kT2<br />
t0<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
, dini))<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2) αs(d1)<br />
dini))<br />
αs(dini) αs(dini)<br />
kT3<br />
54
le of <strong>the</strong> procedure<br />
• Apply <strong>the</strong> required Sudakov suppression<br />
simulate pp → W + jets.<br />
k (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
k momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
CKKW matching<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
analytically, using <strong>the</strong> best available (NLL) Sudakovs.<br />
• Per<strong>for</strong>m “truncated showering”: Run <strong>the</strong> parton shower<br />
starting at t0, but <strong>for</strong>bid any showers above <strong>the</strong> cutoff scale tcut.<br />
kT1<br />
x<br />
x<br />
kT2<br />
t0<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
, dini))<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2) αs(d1)<br />
dini))<br />
αs(dini) αs(dini)<br />
kT3<br />
54
le of <strong>the</strong> procedure<br />
• Apply <strong>the</strong> required Sudakov suppression<br />
simulate pp → W + jets.<br />
k (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
k momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
ter <strong>the</strong> event using <strong>the</strong><br />
variant kT clustering<br />
, to get nodes d1, d2, d3 as<br />
ly <strong>the</strong> αs and Sudakov<br />
CKKW matching<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
analytically, using <strong>the</strong> best available (NLL) Sudakovs.<br />
• Per<strong>for</strong>m “truncated showering”: Run <strong>the</strong> parton shower<br />
starting at t0, but <strong>for</strong>bid any showers above <strong>the</strong> cutoff scale tcut.<br />
kT1<br />
x<br />
✓ Best <strong>the</strong>oretical treatment of matrix element<br />
- Requires dedicated PS implementation<br />
x<br />
t0<br />
[Catani, Krauss, Kuhn, Webber 2001]<br />
[Krauss 2002]<br />
- Mismatch between analytical kT2 Sudakov and (non-NLL) shower<br />
• Implemented in Sherpa (v. 1.1)<br />
, dini))<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2) αs(d1)<br />
dini))<br />
αs(dini) αs(dini)<br />
kT3<br />
54
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
CKKW-L matching<br />
αs(d1)<br />
αs(dini)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
CKKW-L matching<br />
αs(d1)<br />
αs(dini)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
CKKW-L matching<br />
αs(d1)<br />
αs(dini)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
• Veto <strong>the</strong> event if any shower is harder than <strong>the</strong> clustering scale<br />
<strong>for</strong> <strong>the</strong> next step (or tcut, if last step)<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
7 / 23<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
• Veto <strong>the</strong> event if any shower is harder than <strong>the</strong> clustering scale<br />
<strong>for</strong> <strong>the</strong> next step (or tcut, if last step)<br />
• Keep any shower emissions that are softer 7 than / 23 <strong>the</strong> clustering<br />
scale <strong>for</strong> <strong>the</strong> next step<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
55
ick (according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
Wd ¯d event<br />
ick momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u¯d→Wd ¯d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯d (x2, dini)<br />
ster <strong>the</strong> event using <strong>the</strong><br />
invariant kT clustering<br />
e, to get nodes d1, d2, d3 as<br />
ply <strong>the</strong> αs and Sudakov<br />
• Cluster back to “parton shower history”<br />
αs(dini) αs(dini)<br />
• Per<strong>for</strong>m showering step-by-step <strong>for</strong> each step in <strong>the</strong> parton<br />
shower history, starting from <strong>the</strong> clustering scale <strong>for</strong> that step<br />
✓• Automatic Veto <strong>the</strong> event agreement if any shower between is Sudakov harder than and <strong>the</strong> shower clustering scale<br />
<strong>for</strong> <strong>the</strong> next step (or tcut, if last step)<br />
- Requires dedicated PS implementation<br />
• Keep any shower emissions that are softer 7 than / 23 <strong>the</strong> clustering<br />
➡ Need multiple implementations to compare between showers<br />
scale <strong>for</strong> <strong>the</strong> next step<br />
• Implemented in Ariadne, Sherpa (v. 1.2), and Pythia 8<br />
3, dini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
CKKW-L matching<br />
αs(d1)<br />
pply initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
MW , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
eto all emissions above dini (in both initial- and final state showers).<br />
kT2<br />
kT1<br />
t0<br />
[Lönnblad 2002]<br />
[Hoeche et al. 2009]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
55
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
αs(d1)<br />
αs(dini)<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
t0<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
56
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
αs(d1)<br />
αs(dini)<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
t0<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
• The simplest way to do <strong>the</strong> Sudakov suppression is to run <strong>the</strong><br />
shower on <strong>the</strong> event, starting from t0!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
56
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
αs(dini)<br />
αs(d1)<br />
αs(dini)<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
kT2<br />
t0<br />
kT3<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
• The simplest way to do <strong>the</strong> Sudakov suppression is to run <strong>the</strong><br />
shower on <strong>the</strong> event, starting from t0!<br />
56
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
• Per<strong>for</strong>m jet clustering αs(d1) after PS - if hardest jet kT1 > tcut or<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
<strong>the</strong>re are jets αs(dini) not matched αs(dini) to partons, reject <strong>the</strong> event<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
kT2<br />
t0<br />
kT3<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
• The simplest way to do <strong>the</strong> Sudakov suppression is to run <strong>the</strong><br />
shower on <strong>the</strong> event, starting from t0!<br />
56
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
• Per<strong>for</strong>m jet clustering αs(d1) after PS - if hardest jet kT1 > tcut or<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
<strong>the</strong>re are jets αs(dini) not matched αs(dini) to partons, reject <strong>the</strong> event<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
kT2<br />
t0<br />
kT3<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
(∆Iq(tcut,t0)) 2 (∆q(tcut,t0)) 2<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
• The simplest way to do <strong>the</strong> Sudakov suppression is to run <strong>the</strong><br />
shower on <strong>the</strong> event, starting from t0!<br />
• The resulting Sudakov suppression from <strong>the</strong> procedure is<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
which turns out to be a good enough approximation of <strong>the</strong><br />
correct expression<br />
56
of <strong>the</strong> procedure<br />
simulate pp → W + jets.<br />
MLM matching<br />
(according to <strong>the</strong> relative cross-section of <strong>the</strong> processes) a<br />
d ¯d event<br />
momenta according to <strong>the</strong> pdf-weighted matrix element<br />
|M u ¯ d→Wd ¯ d (x1, x2, αs(dini))| 2 fu(x1, dini)f¯ d (x2, dini)<br />
r <strong>the</strong> event using <strong>the</strong><br />
ariant kT clustering<br />
o get nodes d1, d2, d3 as<br />
<strong>the</strong> αs and Sudakov<br />
• Per<strong>for</strong>m jet clustering αs(d1) after PS - if hardest jet kT1 > tcut or<br />
ini)) 2 ∆g (d2, dini)<br />
∆g (d1, dini) (∆q(d1,<br />
2 αs(d2)<br />
dini))<br />
<strong>the</strong>re are jets αs(dini) not matched αs(dini) to partons, reject <strong>the</strong> event<br />
ly initial-state radiation <strong>for</strong> <strong>the</strong> incoming u and ¯d starting at<br />
W , and final-state radiation <strong>for</strong> <strong>the</strong> outgoing d and ¯d starting at d2,<br />
all emissions above dini (in both initial- and final state showers).<br />
kT1<br />
kT2<br />
t0<br />
kT3<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> 7 / 23<strong>simulation</strong><br />
<strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
kT4<br />
(∆Iq(tcut,t0)) 2 (∆q(tcut,t0)) 2<br />
[M.L. Mangano, ~2002, 2007]<br />
[J.A. et al 2007, 2008]<br />
• The simplest way to do <strong>the</strong> Sudakov suppression is to run <strong>the</strong><br />
shower on <strong>the</strong> event, starting from t0!<br />
✓ Simplest available scheme<br />
✓• Allows The resulting matching Sudakov with any suppression shower, without from <strong>the</strong> modification procedure is<br />
(∆Iq(tcut,t0)) 2 ∆g(t2,t1)(∆q(tcut,t2)) 2<br />
➡ Sudakov<br />
which turns<br />
suppression<br />
out to be<br />
not<br />
a good<br />
exact,<br />
enough<br />
minor<br />
approximation<br />
mismatch with<br />
of<br />
shower<br />
<strong>the</strong><br />
• Implemented correct expression in AlpGen, HELAC, MadGraph<br />
56
•<br />
•<br />
•<br />
Highest multiplicity sample<br />
In <strong>the</strong> previous, assumed we can simulate all parton<br />
multiplicities by <strong>the</strong> ME<br />
In practice, we can only do limited number of final-state<br />
partons with matrix element (up to 4-5 or so)<br />
For <strong>the</strong> highest jet multiplicity that we generate with <strong>the</strong><br />
matrix element, we need to allow additional jets above<br />
<strong>the</strong> matching scale tcut, since we will o<strong>the</strong>rwise not get a<br />
jet-inclusive description – but still can’t allow PS radiation<br />
harder than <strong>the</strong> ME partons<br />
➡ Need to replace tcut by <strong>the</strong> clustering scale <strong>for</strong> <strong>the</strong> softest<br />
ME parton <strong>for</strong> <strong>the</strong> highest multiplicity<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
57
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Matching scale<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Matching scale<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Matching scale<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Matching scale<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
•<br />
•<br />
•<br />
Back to <strong>the</strong> “matching goal”<br />
Regularization of matrix element divergence<br />
Correction of <strong>the</strong> parton shower <strong>for</strong> large momenta<br />
Smooth jet distributions<br />
Matrix element<br />
Parton shower<br />
Matching scale<br />
Desired curve<br />
2nd QCD radiation jet in<br />
top pair production at<br />
<strong>the</strong> <strong>LHC</strong>, using<br />
MadGraph + Pythia<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
58
Summary of Matching Procedure<br />
1. Generate ME events (with different parton multiplicities)<br />
using parton-level cuts (pT ME /ΔR or kT ME )<br />
2. Cluster each event and reweight αs and PDFs based on <strong>the</strong><br />
scales in <strong>the</strong> clustering vertices<br />
3. Apply Sudakov factors to account <strong>for</strong> <strong>the</strong> required nonradiation<br />
above clustering cutoff scale and generate parton<br />
shower emissions below clustering cutoff:<br />
a. (CKKW) Analytical Sudakovs + truncated showers<br />
b. (CKKW-L) Sudakovs from truncated showers<br />
c. (MLM) Sudakovs from reclustered shower emissions<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
59
[pb]<br />
T<br />
)dE<br />
T<br />
(d/dE<br />
<br />
2<br />
10<br />
min<br />
ET<br />
10<br />
1<br />
-1<br />
10<br />
-2<br />
10<br />
Comparing to experiment: W+jets<br />
(We)<br />
+ n<br />
jets<br />
st<br />
1 jet<br />
nd<br />
2 jet<br />
rd<br />
3 jet<br />
th<br />
4 jet<br />
CDF Data<br />
CDF Run II Preliminary<br />
<br />
-1<br />
dL = 320 pb<br />
e<br />
W kin: E 20[GeV];<br />
| |<br />
1.1<br />
0 50 100 150 200<br />
min<br />
Jet Transverse Energy (E ) [GeV]<br />
T<br />
W<br />
T<br />
e<br />
<br />
T<br />
M 20[GeV/c<br />
]; E 30[GeV]<br />
Jets: JetClu R=0.4; | | 30 GeV<br />
1 2 3 4<br />
inclusive jet multiplicity, n<br />
60
MLM matching schemes in MadGraph<br />
[J.A. et al (2007, 2008)]<br />
[J.A. et al (2011)]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
61
MLM matching schemes in MadGraph<br />
In MadGraph, <strong>the</strong>re are 3 different MLM-type matching<br />
schemes differing in how to divide ME vs. PS regions:<br />
[J.A. et al (2007, 2008)]<br />
[J.A. et al (2011)]<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
61
MLM matching schemes in MadGraph<br />
[J.A. et al (2007, 2008)]<br />
[J.A. et al (2011)]<br />
In MadGraph, <strong>the</strong>re are 3 different MLM-type matching<br />
schemes differing in how to divide ME vs. PS regions:<br />
a. Cone jet MLM scheme:<br />
- Use cuts in pT (pT ME )and ΔR between partons in ME<br />
- Cluster events after parton shower using a cone jet<br />
algorithm with <strong>the</strong> same ΔR and pT match > pT ME<br />
- Keep event if all jets are matched to ME partons (i.e., all<br />
ME partons are within ΔR of a jet)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
61
MLM matching schemes in MadGraph<br />
[J.A. et al (2007, 2008)]<br />
[J.A. et al (2011)]<br />
In MadGraph, <strong>the</strong>re are 3 different MLM-type matching<br />
schemes differing in how to divide ME vs. PS regions:<br />
a. Cone jet MLM scheme:<br />
- Use cuts in pT (pT ME )and ΔR between partons in ME<br />
- Cluster events after parton shower using a cone jet<br />
algorithm with <strong>the</strong> same ΔR and pT match > pT ME<br />
- Keep event if all jets are matched to ME partons (i.e., all<br />
ME partons are within ΔR of a jet)<br />
b. kT-jet MLM scheme:<br />
- Use cut in <strong>the</strong> Durham kT in ME<br />
- Cluster events after parton shower using <strong>the</strong> same kT<br />
clustering algorithm into kT jets with kT match > kT ME<br />
- Keep event if all jets are matched to ME partons<br />
(i.e., all partons are within kT match to a jet)<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
61
MLM matching schemes in MadGraph<br />
c. Shower-kT scheme:<br />
- Use cut in <strong>the</strong> Durham kT in ME<br />
- After parton shower, get in<strong>for</strong>mation from <strong>the</strong> PS<br />
generator about <strong>the</strong> kT PS of <strong>the</strong> hardest shower<br />
emission<br />
- Keep event if kT PS < kT match<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
62
How to do matching in MadGraph+Pythia<br />
Example: Simulation of pp→W with 0, 1, 2 jets<br />
mg5> generate p p > w+, w+ > l+ vl @0<br />
mg5> add process p p > w+ j, w+ > l+ vl @1<br />
mg5> add process p p > w+ j j, w+ > l+ vl @2<br />
mg5> output<br />
In run_card.dat:<br />
…<br />
…<br />
…<br />
1 = ickkw<br />
0 = ptj<br />
15 = xqcut<br />
(com<strong>for</strong>table on a laptop)<br />
Matching on<br />
kT matching scale<br />
Matching automatically done when run through<br />
MadEvent and Pythia!<br />
No cone matching<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
63
How to do matching in MadGraph+Pythia<br />
• By default, kT-MLM matching is run if xqcut > 0, with <strong>the</strong><br />
matching scale QCUT = max(xqcut*1.4, xqcut+10)<br />
• For shower-kT, by default QCUT = xqcut<br />
• If you want to change <strong>the</strong> Pythia setting <strong>for</strong> matching<br />
scale or switch to shower-kT matching:<br />
In pythia_card.dat:<br />
…<br />
! This sets <strong>the</strong> matching scale, needs to be > xqcut<br />
QCUT = 30<br />
! This switches from kT-MLM to shower-kT matching<br />
! Note that MSTP(81)>=20 needed (pT-ordered shower)<br />
SHOWERKT = T<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
64
How to do validate <strong>the</strong> matching<br />
• The matched cross section is found at <strong>the</strong> end of <strong>the</strong><br />
Pythia log file<br />
• The matched cross section (<strong>for</strong> X+0,1,... jets) should be<br />
close to <strong>the</strong> unmatched cross section <strong>for</strong> <strong>the</strong> 0-jet sample<br />
• The matching scale (QCUT) should typically be chosen<br />
around 1/6-1/2 x hard scale (so xqcut correspondingly<br />
lower)<br />
• The differential jet rate plots should be smooth<br />
• When QCUT is varied (within <strong>the</strong> region of validity), <strong>the</strong><br />
matched cross section or differential jet rates should not<br />
vary significantly<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
65
Matching validation<br />
W+jets production at <strong>the</strong> Tevatron <strong>for</strong> MadGraph+Pythia<br />
(kT-jet MLM scheme, q 2 -ordered Pythia showers)<br />
Q match = 10 GeV Q match = 30 GeV<br />
log(Differential jet rate <strong>for</strong> 1 → 2 radiated jets ~ pT(2nd jet))<br />
Jet distributions smooth, and stable when we vary <strong>the</strong> matching scale!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
66
Summary of Lecture 1<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
67
•<br />
Summary of Lecture 1<br />
Despite <strong>the</strong> apparent enormous complexity of <strong>simulation</strong><br />
of complete collider events, nature has kindly allowed us<br />
to factorize <strong>the</strong> <strong>simulation</strong> into separate steps<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
67
•<br />
•<br />
Summary of Lecture 1<br />
Despite <strong>the</strong> apparent enormous complexity of <strong>simulation</strong><br />
of complete collider events, nature has kindly allowed us<br />
to factorize <strong>the</strong> <strong>simulation</strong> into separate steps<br />
The <strong>Monte</strong> <strong>Carlo</strong> method allows us to step-by-step<br />
simulate hard scattering, parton shower, particle decays,<br />
hadronization, and underlying event<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
67
•<br />
•<br />
•<br />
Summary of Lecture 1<br />
Despite <strong>the</strong> apparent enormous complexity of <strong>simulation</strong><br />
of complete collider events, nature has kindly allowed us<br />
to factorize <strong>the</strong> <strong>simulation</strong> into separate steps<br />
The <strong>Monte</strong> <strong>Carlo</strong> method allows us to step-by-step<br />
simulate hard scattering, parton shower, particle decays,<br />
hadronization, and underlying event<br />
Jet matching between matrix elements and parton<br />
showers gives crucial improvement of <strong>simulation</strong> of<br />
background as well as signal processes<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
67
•<br />
•<br />
•<br />
•<br />
Summary of Lecture 1<br />
Despite <strong>the</strong> apparent enormous complexity of <strong>simulation</strong><br />
of complete collider events, nature has kindly allowed us<br />
to factorize <strong>the</strong> <strong>simulation</strong> into separate steps<br />
The <strong>Monte</strong> <strong>Carlo</strong> method allows us to step-by-step<br />
simulate hard scattering, parton shower, particle decays,<br />
hadronization, and underlying event<br />
Jet matching between matrix elements and parton<br />
showers gives crucial improvement of <strong>simulation</strong> of<br />
background as well as signal processes<br />
Next lecture: Next-to-leading order <strong>simulation</strong>s and<br />
workflow <strong>for</strong> New Physics <strong>simulation</strong>!<br />
PreSUSY, Beijing, August 8-11, 2012 <strong>Monte</strong> <strong>Carlo</strong> <strong>simulation</strong> <strong>for</strong> <strong>the</strong> <strong>LHC</strong> Johan Alwall<br />
67