rivet as an experiment-theory interface for the heavy-ion
TRANSCRIPT
Rivet as an experiment-theory interfacefor the heavy-ion community
Przemek Karczmarczykfor the ALICE Collaboration
MPI@LHC, 18.11.2019
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 1 / 16
Introduction
Introduction
Initial idea
Create an interface between experimental data and theoreticalpredictions for the heavy-ion community
� What do we have?
I Analyses (usually with finite acceptance, non-trivial cuts, non-trivialalgorithms of extracting physics objects, etc.)
I Data from different experiments published in papers (values usuallypublished in HepData)
I Large number of different Monte Carlo event generators
� What do we want?
I Test MC models by comparing their results to the experiment → RivetI Store the results in an easily accessible repository → MCPLOTS
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 2 / 16
Rivet
Rivet
� Rivet: Robust Independent Validation of Experiment and TheoryI Generator-independent analysis framework, arXiv:1003.0694I System for validation and tuning of event generatorsI Data & analysis algorithms preservation
� Rivet features:I Read input from HepMC format (file or FIFO)I Provides a large set of experimental analyses (currently over 400...)I Run one (or more) analyses on the input data from generatorI Fast and direct comparison between exp. data and generators
Example of Rivet results
Analysis coverage, source:https://rivet.hepforge.org/rivet-coverage
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 3 / 16
Rivet
Rivet analysis
� Analyses as C++ classes
� To be written by users (analysers)
� Unique analysis identifier
I Consist of experiment name, year of publication, and Inspire ID codeI e.g. ALICE 2015 I1357424
� 3 core methods:
I init(): histograms booking, variables initialization...I analyze(...): main event loop, events are analysed and histograms are
filled hereI finalize(): histograms scaling, divisions for the final data comparison
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 4 / 16
Rivet
Rivet for heavy-ions
� Rivet originally developed for pp/pp̄ and e+e− physics
� But recently this has changed
I More interest from the heavy-ion communityI Possibility to use Rivet as a tool for HI analyses
� Most common HI features covered in Rivet 2.7 (see link)
� Rivet 2.7 released in 2019 with:
I Centrality calibration & selection using analysis optionsI Comparing and merging Rivet runs using reentrant finalize (e.g. for
AA/pp ratio)I Flow observables, event mixing frameworkI A set of 20+ analyses using heavy-ion features
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 5 / 16
Rivet Centrality framework
Centrality framework in Rivet
� Centrality framework: allows an analysis to cut on percentiles ofsingle event quantities preloaded to that analysis
� Consist of:I A set of calibration analysesI Preloading calibration files as an input to the next runI Options for centrality to select the type of calibration to useI Centrality projection that allows to access centrality value
� Analysis options allow selection of centrality calibration method1 Calibration histogram from reference data2 Generated calibration histogram3 Impact parameter calibration histogram4 User-defined calibration histogram5 Generated centrality (available only with HepMC3)
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 6 / 16
Rivet Centrality framework
Centrality calibration
� Calibration analysis used to produce calibration files
� Example analysis: ALICE 2015 PBPBCentrality
MC
0 5 10 15 20
10−3
10−2
10−1
b (fm)
(1/
Nev
t)d
N/
db
Impact parameter distribution
b
bb b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
b b b b b b b b b b bb b b b b b b b b b b b b b b b b
bb b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
b b b b bb b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
b b b b b bb b b b b b b b b b b b b
bbb b b b b b b bbb b b b b bb b b
b bbb
b b bbb b b b
b DataMC
0 5.0 ·103 1.0 ·104 1.5 ·104 2.0 ·104 2.5 ·104 3.0 ·104 3.5 ·104
10−6
10−5
10−4
10−3
10−2
10−1
1
V0M multiplicity distribution
� New preload flag to preload histograms to the Rivet runs
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 7 / 16
Rivet Centrality framework
Centrality selection with analysis options
� Analysis options
I Possible to select by user at runtimeI Analysis can be run with different optionsI Centrality value accessible inside an analysis
b
b
b
b
b
b
bb
b
b DataMC [cent=IMP]MC [cent=GEN]
0
200
400
600
800
1.0 ·103
1.2 ·103
1.4 ·103
1.6 ·103
1.8 ·103
Nch vs. centrality, Pb–Pb√
sNN = 2.76 TeV
dN
ch/
dη
b b b b b b b b b
0 10 20 30 40 50 60 70 800.50.60.70.80.9
11.11.21.31.4
Centrality [%]
MC
/Dat
a
Example fromALICE 2010 I880049
b b b b b b b b b b b b b b b b b b b b b b b b b b bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb bb b
bb
b DataMC [cent=IMP]MC [cent=GEN]
10−4
10−3
10−2
10−1
1
10 1
10 2
Nch vs. p⊥, Centr = 40 − 60 %,√
sNN = 2.76 TeV
1/N
evt1
/(2
πp ⊥
)(d
2 Nch)/
(dη
dp ⊥
)[(
GeV
/c)
−2 ]
b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
1 10 10.50.60.70.80.9
11.11.21.31.4
p⊥ [GeV / c]
MC
/Dat
a
Example fromALICE 2012 I1127497
b
b bb
b DataEPOS-LHC
0
0.5
1
1.5
2
2.5
ICP (0-5% / 60-90%) near-side
I CP
(0-5
%/
60-9
0%)
b b b b
3 4 5 6 7 8 9 100.50.60.70.80.9
11.11.21.31.4
pt,assoc [GeV/c]
MC
/Dat
a
Example fromALICE 2012 I930312
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 8 / 16
Rivet Reentrant finalize
Reentrant finalize
� Postprocessing implemented in a form of ’reentrant finalize’ methodI Analyses produce ’RAW’ histograms - saved in the output file before
calling finalize() methodI rivet-merge to call finalize() again, but this time with preloaded files
from previous runs to get final plotsI ’Reentrant’ flag to mark analyses using reentrant finalize method
Nuclear modification factor
RAA = dNAA/dpT<Ncoll>dNpp/dpT
b b b b b b b bb b
b b bb b b
b bb b b
b b b b b b b b b b b b b b bbbbb b b b b b b b
b bb bbbb b
b bbb b b
b b bb
b Dataoutput [cent=IMP]
10−1
1
RAA vs. p⊥, Centr = 0 − 20 %,√
sNN = 2.76 TeV
RA
Ab b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
1 10 10.50.60.70.80.9
11.11.21.31.4
p⊥ [GeV / c]
MC
/Dat
a
ALICE 2012 I1127497 example
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 9 / 16
Rivet Reentrant finalize
HI analyses list
With data:ALICE 2010 I880049 (PbPb): Multiplicity
ALICE 2012 I930312 (PbPb): IAA
ALICE 2012 I1127497 (PbPb): RAA
ALICE 2012 I1126966 (PbPb): π, K, p spectra
ALICE 2013 I1225979 (PbPb): Charged multiplicity
ALICE 2014 I1243865 (PbPb): Multi-strange baryons
ALICE 2014 I1244523 (pPb): Multi-strange baryons
ALICE 2015 PBPBCentrality (PbPb): Calibration
ALICE 2016 I1394676 (PbPb): Charged multiplicity
ALICE 2016 I1419244 (PbPb): Multiparticle correlations (flow)
ALICE 2016 I1471838 (pp): Multi-strange baryons
ALICE 2016 I1507090 (PbPb): Charged multiplicity
ALICE 2016 I1507157 (pp): Angular correlations
ATLAS 2015 I1360290 (PbPb): Charged multiplicity
ATLAS 2015 I1386475 (pPb): Charged multiplicity + spectra
ATLAS PBPB CENTRALITY (PbPb): Calibration
ATLAS pPb Calib (pPb): Calibration
BRAHMS 2004 I647076 (AuAu): π, K, p spectra as function of rapidity
STAR 2016 I1414638 (AuAu): Di-hadron correlations (flow)
CMS 2017 I1471287 (pp): Multiparticle correlations (flow)
LHCF 2016 I1385877 (pPb): Forward pT
Without data:ALICE 2015 PPCentrality: Calibration
BRAHMS 2004 CENTRALITY: Calibration
STAR BES CALIB: Calibration
MC Cent pPb Calib: Calibration example
MC Cent pPb Eta: Calibration example
MC OPTIONS: Analysis options example
MC REENTRANT: Reentrant finalize example
See list of all analyses here
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 10 / 16
MCPLOTS project
MCPLOTS website
� MCPLOTS: a particle physics resource based onvolunteer computing
I The idea: enable anyone to quickly get an idea of howwell a particular model describes various data sets
I Automatize Rivet creation of comparison plotsI Online repository of plots: http://mcplots.cern.ch
[GeV]T
p
)T
dpη)/
(dch
N2)
(dT
pπ 2
ev1/
(N
8−10
7−10
6−10
5−10
4−10
3−10
2−10
1−10
1
10
210
310
410
ALICEHerwig++ (Def)Pythia 6 (Def)Pythia 8 (Def)Sherpa (Def)
900 GeV pp Soft QCD
mcp
lots
.cer
n.ch
4.1
M e
vent
s≥
Riv
et 2
.4.0
,
Herwig++ 2.7.1, Pythia 6.428, Pythia 8.212, Sherpa 2.2.0
ALICE_2010_S8706239
> 0.15 GeV)T
| < 0.8, pη Spectrum (|T
Charged Particle p
[GeV]T
p1 10
Rat
io to
ALI
CE
1
2
0.5
+20%
-20%
+10%
-10%
2×
-50%
+50%
[GeV]T
p
Tdpη
/dσ d
Tpπ 1
/2ev
1/N
9−10
8−10
7−10
6−10
5−10
4−10
3−10
2−10
1−10
1
10ATLASHerwig++ (Def)Pythia 6 (Def)Pythia 8 (Def)Sherpa (Def)
7000 GeV pp Soft QCD
mcp
lots
.cer
n.ch
4.9
M e
vent
s≥
Riv
et 2
.4.0
,
Herwig++ 2.7.1, Pythia 6.428, Pythia 8.212, Sherpa 2.2.0
ATLAS_2010_S8918562
> 2.5 GeV)T
> 1, pch
Spectrum (NT
Charged Particle p
[GeV]T
p10
Rat
io to
AT
LAS
1
2
0.5
+20%
-20%
+10%
-10%
2×
-50%
+50%
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 11 / 16
MCPLOTS project
MCPLOTS for heavy-ions
� Originally provided results for pp and ee, but not for heavy-ions� New heavy-ion version of MCPLOTS
I Uses new Rivet version as a baselineI Enables to run analyses that use heavy-ion features described
� Integration ongoingI Preview version available here: http://mcplots-alice.cern.ch
Centrality [%]
η /
dch
dN
210
310
410
ALICEEpos (LHC)
2760 GeV PbPb Soft QCD
mcp
lots
.cer
n.ch
(ar
Xiv
:130
6.34
36)
10k
eve
nts
≥R
ivet
2.7
.0,
Epos 1.99.crmc.1.7.0-hi
ALICE_2010_I880049
> 0.05 GeV)T
| < 0.5, pη vs centrality (|η/dchdN
Centrality [%]0 20 40 60 80
Rat
io to
ALI
CE
1
2
0.5
2
0.5
1
η
η /
dch
dN
400
600
800
1000
1200
1400
1600
1800
2000
2200
2400
ALICEEpos (LHC)
2760 GeV PbPb Soft QCD
mcp
lots
.cer
n.ch
(ar
Xiv
:130
6.34
36)
10k
eve
nts
≥R
ivet
2.7
.0,
Epos 1.99.crmc.1.7.0-hi
ALICE_2013_I1225979 (UNVALIDATED)
| < 5.6, 0-5%)η distribution (|ηCharged particle
η5− 0 5
Rat
io to
ALI
CE
1
2
0.5
2
0.5
1
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 12 / 16
Summary
Summary
� Rivet
I Rivet 2.7 containing first heavy-ion features releasedI Already possible to calibrate analysis, use centrality, and obtain ratios
of AA/ppI Several heavy-ion analyses already included in the release
� MCPLOTS
I Adapted to the heavy-ion features available in the recent Rivet releaseI In the process of integration with the main MCPLOTS repository
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 13 / 16
Summary
Summary
� Rivet
I Rivet 2.7 containing first heavy-ion features releasedI Already possible to calibrate analysis, use centrality, and obtain ratios
of AA/ppI Several heavy-ion analyses already included in the release
� MCPLOTS
I Adapted to the heavy-ion features available in the recent Rivet releaseI In the process of integration with the main MCPLOTS repository
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 13 / 16
Summary
What next?
� Continuation of development of the frameworksI More heavy-ion features (e.g. fitting)I More analyses coming in
� Integration of more HI generatorsI Currently available: Epos (LHC), Jewel, Pythia, ...I We need more of them!
� First framework is there: now we need more people working onanalyses integration
HI analysis coverage, source:https://rivet.hepforge.org/rivet-coverage-heavyiononly
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 14 / 16
Summary
What next?
� Continuation of development of the frameworksI More heavy-ion features (e.g. fitting)I More analyses coming in
� Integration of more HI generatorsI Currently available: Epos (LHC), Jewel, Pythia, ...I We need more of them!
� First framework is there: now we need more people working onanalyses integration
HI analysis coverage, source:https://rivet.hepforge.org/rivet-coverage-heavyiononly
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 14 / 16
Summary
What next?
� Continuation of development of the frameworksI More heavy-ion features (e.g. fitting)I More analyses coming in
� Integration of more HI generatorsI Currently available: Epos (LHC), Jewel, Pythia, ...I We need more of them!
� First framework is there: now we need more people working onanalyses integration
HI analysis coverage, source:https://rivet.hepforge.org/rivet-coverage-heavyiononly
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 14 / 16
Summary
References
� Rivet webpage: https://rivet.hepforge.org/
� Rivet user manual: https://arxiv.org/pdf/1003.0694.pdf
� Rivet source code: https://phab.hepforge.org/source/rivethg/
� MCPLOTS source code: https://gitlab.cern.ch/MCPLOTS/mcplots
� MCPLOTS webpage: http://mcplots.cern.ch/
� MCPLOTS HI development webpage: http://mcplots-alice.cern.ch
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 15 / 16
Summary
Thank you!
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 16 / 16
Summary
Backup
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 17 / 16
Summary
More about Rivet
� Rivet features
I Provides a large set of experimental analyses (currently over 400...)useful for MC generator development, validation, and tuning
I Analyses correspond to the actual paper resultsI Analysis algorithms in object-oriented C++I Analysis code separated → easy way to add a new oneI Using the HepMC event format → independent of MC generatorsI Plotting based on YODA framework
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 18 / 16
Summary
HepMC & YODA
� HepMC:I Standarized event record library for High Energy Physics Monte Carlo
generators and simulationI A direct output of most of modern generatorsI File format contains information about final state particlesI Why? We need an interface between generator and analysis
� YODA (Yet more Objects for Data Analysis):
I A histogramming toolkit developed as a lightweightcommon system for MC event generator validationanalyses
I The core histogramming system in Rivet YODA logoP. Karczmarczyk Rivet for heavy-ions 18.11.2019 19 / 16
Summary
HI analysis chain
1 Calibration run:rivet -a ALICE_Calibration /path/to/PbPb.hepmc
2 Run with preloaded calibration file for PbPb beam:rivet -p /path/to/calib.yoda -a ALICE_2012_I123456:cent=IMP /path/to/PbPb.hepmc
3 Regular run for pp beam:rivet -a ALICE_2012_I123456:cent=IMP /path/to/pp.hepmc
4 Postprocessing with reentrant finalize:rivet -merge PbPb.yoda pp.yoda -o final.yoda
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 20 / 16
Summary
RAA analysis
b b b b b b b b b b b b b b b b b b b b b b b b b b bbbbbbbbbbbb
bbbbbbbbbbbbbbbbbbbb bbb bbbb
b Dataoutput [cent=IMP]
10−3
10−2
10−1
1
10 1
10 2
10 3
Nch vs. p⊥, Centr = 0 − 10 %,√
sNN = 2.76 TeV
1/N
evt1
/(2
πp ⊥
)(d
2 Nch)/
(dη
dp ⊥
)[(
GeV
/c)
−2 ]
b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b
1 10 10.50.60.70.80.9
11.11.21.31.4
p⊥ [GeV / c]
MC
/Dat
a
(a) PbPb results
output [cent=IMP]
1 10 1
10−5
10−4
10−3
10−2
10−1
1
Nch vs. p⊥, Centr = 0 − 10 %,√
sNN = 2.76 TeV
p⊥ [GeV / c]
Nev
t1/(2
πp ⊥
)(d
2 Nch)/
(dη
dp ⊥
)[(
GeV
/c)
−2 ]
(b) pp reference
� Impact parameter distribution used for calibration� No HepMC data for pp reference� JEWEL does not provide Ncoll - that’s why the difference between
MC and experimental dataP. Karczmarczyk Rivet for heavy-ions 18.11.2019 21 / 16
Summary
RAA analysis
Nuclear modification factor
RAA = dNAA/dpT<Ncoll>dNpp/dpT
〈Ncoll〉 - average number of nucleon-nucleon collisions in a HI collisiondNAA/dpT - transverse momentum distribution in HI collision
dNpp/dpT - transverse momentum distribution in pp collision
� RAA tells how heavy-ion events looks like in scale of pp collisions
� Both pp and heavy-ion collisions are required to calculate RAA
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 22 / 16
Summary
MCPLOTS project
� The mcplots.cern.ch web siteI Simple online repository of plots made using RivetI with MC generators and comparisons to the experimental dataI Originally provides results for pp and ee, but not for heavy-ions
� Relies on:I The HEPDATA database of experimental resultsI The Rivet Monte Carlo analysis toolI MC event generators (currently implemented: ALPGEN, EPOS-LHC,
HERWIG++, PHOJET, PYTHIA 6, PYTHIA 8, SHERPA, VINCIA)I The LHC@HOME 2.0 framework for volunteer cloud computing
P. Karczmarczyk Rivet for heavy-ions 18.11.2019 23 / 16