Error: (1) can't find atlaslogo.png at /Sandbox.SimoneAmorosoSandbox

Physics Modelling Group (PMG)


Frank Siegert and Simone Amoroso

Mail atlas-phys-pmg-conveners@cernNOSPAMPLEASE.ch
Error: (1) can't find PMG_Structure.png at /Sandbox.SimoneAmorosoSandbox

PMG has 4 new subgroups!
Look at #Group_Overview.
This Twiki is currently being restructured accordingly.

The respository of sample information for analyses:

Quick links:

PMG mandate

The Physics Modelling Group is responsible for the development and validation of MC generators and samples for analyses and for the final approval of general MC requests from all physics groups (in tight collaboration with Physics Coordinators). The group is also responsible for the setting of common recipes for theoretical systematic uncertainties and possibly needed re-weighting procedure needed to correct the MC modelling. Each PA and CP group is represented by MC production contacts (responsible for the request and production of samples within their group) and Physics contacts that follow closely needs from their group.

Group Overview

Indico for meetings: Physics Modelling Group

The organisation of the PMG is shown in the chart at the top of the page. It is split into four sub-groups: Generator Infrastructure and Tools, Weak Boson Processes, Top Processes and Jet and Photon Processes. In addition, there are several activity groups: MC Software, MC Validation, MC JobOptions and Rivet support. More information about all the quick links below and the subsections below.

Sub-group/activity group TWiki & contact people Mailing list
Generator Infrastructure and Tools PmgGeneratorInfrastructureAndTools Conveners: Louie Dartmoor Corpe & Shu Li
Mail atlas-phys-pmg-git-conveners@cernNOSPAMPLEASE.ch
Mail atlas-phys-pmg-git@cernNOSPAMPLEASE.ch
Weak Boson Processes PmgWeakBosonProcesses Conveners: Matthew Gignac & Lailin Xu
Mail atlas-phys-pmg-weak-boson-conveners@cernNOSPAMPLEASE.ch
Mail atlas-phys-pmg-weak-boson@cernNOSPAMPLEASE.ch
Top Processes PmgTopProcesses Conveners: Maria Moreno Llacer & Ian Connelly
Mail atlas-phys-pmg-top-conveners@cernNOSPAMPLEASE.ch
Mail atlas-phys-pmg-top@cernNOSPAMPLEASE.ch
Jet & Photon Processes PmgJetAndPhotonProcesses Conveners: Ana Cueto & Nishu Nishu
Mail atlas-phys-pmg-jet-photon-conveners@cernNOSPAMPLEASE.ch
Mail atlas-phys-pmg-jet-photon@cernNOSPAMPLEASE.ch
MC Validation GeneratorValidationPage Coordinators: Harinder Singh Bawa, Deepak Kar
Mail atlas-phys-pmg-mcvalid-conveners@cernNOSPAMPLEASE.ch
MC Software PmgMcSoftware Coordinators: Ewelina Lobodzinska, Chris Hays
Mail atlas-simulation-generators-software@cernNOSPAMPLEASE.ch
MC JobOptions MC15JobOptions Coordinators: Cyril Becot, Eva Bouhova-Thacker, Joany Manjarres
Mail atlas-phys-mcprod-jo@cernNOSPAMPLEASE.ch
Rivet analysis support Rivet Usage Contact: Christian Gutschow
Mail atlas-phys-pmg-rivet@cernNOSPAMPLEASE.ch
PMG MC and Physics Contacts PmgContacts    
Central Page CentralMC15ProductionList Mail atlas-phys-pmg-centralpage-contacts@cernNOSPAMPLEASE.ch  

PMG Public Results

ATLAS public results from the Physics Modelling Group

Uncovered and Ongoing Tasks

Open Tasks

The following table contains a list open topics that need person power in PMG. The experience and time required are stated along with the urgency. Many of these topic could be qualification tasks, but people just interested in contributing are very much encouraged too and OTP will be provided as appropriate. More details on those relevant for qualification tasks can be found in the following JIRA page. If you are interested in contributing to any of these topics please get in contact with the PMG conveners: Mail atlas-phys-pmg-conveners@cernNOSPAMPLEASE.ch.

Error: Unsupported content type: application/octet-stream (Must be text/html, text/plain or text/css)

Ongoing Tasks

Error: Unsupported content type: application/octet-stream (Must be text/html, text/plain or text/css)

Finished Tasks

Error: Unsupported content type: application/octet-stream (Must be text/html, text/plain or text/css)

PMG Recommended/Supported tools

Truth Weights Tool

This tool is designed as a plugin in your AthAnalysis code to correctly retrieve the matrix-element weights from the metadata and access them by name. This means you no longer need to hard-code which weight index corresponds to alpha_S, PDF, scale etc variations.. https://twiki.cern.ch/twiki/bin/view/AtlasProtected/PmgEventWeights

Systematics/Correlations Tool

This tool is designed to help analysers process matrix-element weighs into systematic variations. Also the tool helps to handle bin-bin correlations when evaluating goodness of fit between data and MC. (Uses the Truth Weights Tool) https://twiki.cern.ch/twiki/bin/view/AtlasProtected/PMGSystematicsTool

Cross-Section Tool

The PMG provides a PMGCrossSectionTool for extracting the cross sections, filter efficiencies and k-factors in your analysis framework. This reads from data file that contains all cross sections, filter efficiencies and k-factors as maintained by the PMG.

LaTeX Snippets and References

More information here.

Theory Prediction Guidelines

Treatment of theory predictions in ATLAS: CDS link

Sample information

Baseline samples

For baseline sample recommendations please see the Focus Group pages linked below. These provide information on what samples are available and give guidance which samples are most appropriate for different regions of phase space. Systematic uncertainty recommendations are also provided.

Finding samples

This section hopefully helps you to find the samples needed for your analysis. If anything is unclear please contact the PMG conveners(Mail atlas-phys-pmg-conveners@cernNOSPAMPLEASE.ch)!

Sample status on Focus Group pages

The Focus Groups each have tables that show the production status of the relevant samples. You can find these via the following links:

Boson(+jets) Focus Group:

Top Focus Group:

Multiboson Focus Group:

Rare Top Focus Group:

Central page and PMGCrossSectionTool

The Central Page is the respository for all sample information for analysis.
This includes available samples, known issues, instructions on how to use samples, cross sections and k-factors

If you think the Central Page or AMI has an incorrect value for any of your samples, please report it following the instructions here.

The PMGTools package provides a PMGCrossSectionTool for extracting the cross sections, filter efficiencies and k-factors in your analysis framework. This reads from data file that contains all cross sections, filter efficiencies and k-factors as maintained by the PMG.

Useful links:

MC16 production scheme

The MC16 production scheme is described by the following schematic:

Error: (1) can't find MCProductionStrategy.png at /Sandbox.SimoneAmorosoSandbox

The main message is that for MC16 several datasets need to combined to get the MC sample for several data-taking periods. To find which samples are needed for a given subcampaign the script described below can be used. Note that this will sometimes give you multiple containers for a given subcampaign - this is because in some cases extra merge steps were required for different subsets of the production - but these are independent and can be added together.

Dataset discovery

The dataset naming convention contains a lot of information about how samples are produced via production tags. However, the tags from merging steps are also included which provide little additional information for analysers. Therefore the MC prod team and ProdSys have recently created single-tag (i.e. without merge tags) containers for all EVNT/AOD samples (for e/s/r-tags), This means that you never need to run anything on double-tag containers in the future.

This means that, for example, one should look for the recon.AOD container instead of merge.AOD for final AOD containers.

Here is a an example of what datasets the single-tag containers may contain:

Error: (1) can't find SingleTagContainers.png at /Sandbox.SimoneAmorosoSandbox

For more details on dataset tags for MC16a/c/d/e please look here: AtlasProductionGroupMC16


The PMG recommendations on how to reference samples can be found here (last updated for MC15).

Lots of additional sample information can be found in the following PUB notes. These can be used for more concise references if required/appropriate but are generically useful for getting documentation on the latest generator configurations.

PUB notes

These PUB notes document the latest and greatest ATLAS MC generator configurations and their modelling performance.

2019 PUB notes

Note Glance CDS document Figs & tables Editors
VBS/VBF modelling id=3364 document ATL-PHYS-PUB-2019-004/ M. Shapiro, S. Todt

2018 PUB notes

Note Glance CDS document Figs & tables Editors
ttbar NLO+PS improvements id=3163 document ATL-PHYS-PUB-2018-009 C. Gutschow, A. Knue

2017 PUB notes

Note Glance CDS document Figs & tables Editors
V+jets id=2623 document ATL-PHYS-PUB-2017-006 C. Debenedetti, C. Gutschow, F. Sforza
Multiboson id=2624 document ATL-PHYS-PUB-2017-005 J. Meyer, M.A. Pleier, F. Siegert
Top id=2625 document ATL-PHYS-PUB-2017-007 M. D'Onofrio, D. Hirschbuehl, A. Knue
Colour-reconnection Tunes id=2643 document ATL-PHYS-PUB-2017-008 D. Kar

2016 PUB notes

Note Glance CDS document Figs & tables Editors
V+jets id=1843 document ATL-PHYS-PUB-2016-003 G. Hesketh, P. Thompson
Diboson id=1844 document ATL-PHYS-PUB-2016-002 F. Siegert, M.A. Pleier, C. Gutschow
Top id=1845 document ATL-PHYS-PUB-2016-004 B. Nachman, A. Knue, D. Hirschbuehl‎
ttX id=1846 document ATL-PHYS-PUB-2016-005 J. McFayden, M. Llacer
MC Validation id=1847 document ATL-PHYS-PUB-2016-001 L. Serkin, M. Llacer
ttbar+ccbar id=2023 document ATL-PHYS-PUB-2016-011 J. Katzy, J. Keller, D. Rauch
Top Addendum id=2043 document ATL-PHYS-PUB-2016-016 I. Connelly, M. Llacer, C. Young
A3 Min Bias Tune id=2203 document ATL-PHYS-PUB-2016-017 D. Kar

Systematic uncertainties and reweighting

Systematic uncertainty recipes

Many technical aspects of generator uncertainty recipes are generator-specific rather than process-specific, but there are also many physics aspects of the recipes that are process-specific. This makes it hard to cover the recipes concisely! Here we try to provide links to the an overview of all aspects.


General advice on the non-process-specific aspects of systematic uncertainty recipes are described on PmgSystematicUncertaintyRecipes and direct links given below.

TWiki link Comment
On-the-fly weights Increasingly, systematic variations are being provided in the form of per-event (or on-the-fly) weights. Here we provide general recommendations on how to use these weights.
Normalisation uncertainties General recommendations on how to apply normalisation uncertainties in the case of higher-order cross sections.
QCD scale uncertainties Scale variations are often very important uncertainties. These are usually provided as event weights, the basics of how these are constructed and how to use them are given here.
PDF uncertainties PDF variations are also usually provided as event weights. Typically the LHAPDF tool must be used to extract the correct uncertainty from the weights from the PDF error set, an example using the python implementation of the tool is given.
Parton shower uncertainties Details on both hadronisation and underlying event variations (e.g. Pythia8 vs Herwig7) and shower tune uncertainties (e.g. A14 eigentunes) are given. On-the-fly shower scale variations are under development and my become the norm, but are not yet ready for use in ATLAS.
Negative weights Some general comments on MC statistical uncertaintis in the presence of negative and non-uniform weights.


  • Work in progress, under construction Resummation scale
  • Work in progress, under construction Matching/merging scale


Process-specific systematic uncertainties are the responsibility of the Process Modelling sub-group and the respective Focus Groups. These are linked in the Focus Group section but for convenience are also given here:

Process-specific uncertainties from Focus Groups
Boson+jets Multiboson Work in progress, under construction Multijets Top Rare Top

MC reweighting procedures

Details of PMG reweighting procedures and the tools associated with them can be found at ReweightingProcedure. This provides:

  • A repository of re-weighting procedures as used in physics analyses so far and other methods used in Physics analyses to correct MC predictions
  • Links to special treatments for overlaps and/or MC corrections
  • PMG stance on NNLO ttbar reweighting (this is not recommended)

Process Modelling Subgroup and Focus Groups

Conveners: Jan Kretzschmar & Christian Gutschow
Mail atlas-phys-pmg-processmodelling-conveners@cernNOSPAMPLEASE.ch
The role of the Process Modelling subgroup is to:

  • Overall responsibility for all process-dependent generator configurations
  • Oversee Process Focus Groups
  • Encourage (when appropriate) harmonisation among configurations for different processes
  • Review recommendations for process-specific systematic uncertainties
  • Interact with Physics Analysis groups vis PMG contacts

Focus Groups

Focus Groups Twiki, contact people & mailing list
Boson+jets BosonJetsFocusGroup Matthew Gignac, Carlo Pandini, Evelin Meoni Weds 17:00 atlas-phys-pmg-bosonjets@cernNOSPAMPLEASE.ch
Multiboson MultibosonFocusGroup Louis Helary, Stefanie Todt, Lailin Xu Weds 16:00 atlas-phys-pmg-multi-boson@cernNOSPAMPLEASE.ch
Multijet MultijetFocusGroup Nishu Nishu, Javier Llorente Thur 17:00 atlas-generator-multijets@cernNOSPAMPLEASE.ch
Top TopFocusGroup Monica D'Onofrio, Andrea Knue, Timothee Theveneaux-Pelzer Mon 11:00 atlas-phys-mc-toptuning@cernNOSPAMPLEASE.ch
Rare top RareTopFocusGroup Maria Moreno Llacer, Yichen Li, Jay Howarth Mon 12:00 atlas-phys-pmg-raretop@cernNOSPAMPLEASE.ch

Focus group quick links for higher-order cross sections, baseline samples, systematic uncertainty recipes, known issues and references:

Focus Groups Higher-order cross sections Baseline samples Systematic uncertainties Known issues References
Boson+jets here here here here here
Multibosons here here here here here
Multijets Work in progress, under construction here Work in progress, under construction here Work in progress, under construction
Top here here here here here
Rare Top here here here here here
Higgs PMG          

Physics Groups feedback on MC modelling

Open questions, mismodelling under investigation and items to discuss collected by PMG-contacts

Other MC issues

Generator Development & Tuning Subgroup

Conveners: Spyros Argyropoulos & Shu Li Mail atlas-phys-pmg-gendevtuning-conveners@cernNOSPAMPLEASE.ch
TWiki: PmgGeneratorDevelopmentAndTuning

Supported MC generators

General List of MC generators (still to be updated for MC16-specific generators)

Generator TWiki General mailing list Contact people
Powheg PowhegForATLAS atlas-generators-powhegcontrol@cernNOSPAMPLEASE.ch atlas-generators-powhegcontrol-experts@cernNOSPAMPLEASE.ch
Daniel Hayden, Stefan Richter, Riccardo di Sipio
MadGraph5_aMC@NLO MadGraph5aMCatNLOForAtlas atlas-generators-madgraphcontrol@cernNOSPAMPLEASE.ch atlas-generators-madgraphcontrol-experts@cernNOSPAMPLEASE.ch
Zach Marshall, Josh McFayden, Stefan von Buddenbrock, Hannes Mildner
Sherpa SherpaForAtlas atlas-generators-sherpa@cernNOSPAMPLEASE.ch atlas-generators-sherpa-experts@cernNOSPAMPLEASE.ch
Chris Gutschow, Frank Siegert, Heberth Torres
Pythia8 Pythia8ForAtlas atlas-generators-pythia@cernNOSPAMPLEASE.ch atlas-generators-pythia-experts@cernNOSPAMPLEASE.ch
Andy Buckley, Giancarlo Panizzo, Marjorie Shapiro
Herwig7 Herwig7ForAtlas atlas-generators-herwig7@cernNOSPAMPLEASE.ch atlas-generators-herwig7-experts@cernNOSPAMPLEASE.ch
David Yallup, Ludovic Scyboz, Tetiana Moskalets



Other information relevant to PMG

Running MC generation

Making and monitoring MC requests

Making requests

Most PA/CP groups have their one MC request procedures, but generic instructions for making a MC request can be found here with more detailed information on the request workflow in JIRA here.
Note: The procedure for MC16 requests is slightly more complicated and instructions can be found here.

You can contact the MC production team if you have any questions: atlas-phys-mcprod-team@cernNOSPAMPLEASE.ch

Some additional useful information:

  • The naming convention of jobOption file can be checked by checkjos script.
    • /cvmfs/atlas.cern.ch/repo/sw/Generators/MC15JobOptions/checkjos
    • The physics field characters must be 60 or shorter.
  • Latest caches of AtlasProduction and MCProd for EVGEN production
  • Releases listed in the black list should not be used. Note that it depends on generators.
  • Common parameters of generators

How to recreate job from existing sample

See this FAQ.

Monitoring requests

Instructions for monitoring a MC request can be found on the MCRequestProdSys2 page.

Other useful links

MC Validation

The MC validation Wiki, GeneratorValidationPage, contains the links to the jira tickets used for the validation of the Generators, and instructions for validators and shifters.

MC Software

Pcache purpose of build
MG 2.6.2.atlas2, fixed MGControl (AGENE-1624)
newest Rhadrons; MG 2.6.5 and LHAPDF 6.1.5
xrootd access, jet filters, Sherpa 2.2.6

Rivet Usage

Other information

Old Twiki page


Old Task force pages

V+jets, top, ttV, diboson, photon and higgs samples: TF and details per process

Other PMG-related links

General links

* ATLAS HyperNews, ATLAS egroup
* SVN: Generators, permissions, Set Permissions, AnalysisCommon/PMGTools
* Guidelines for CONF/PUB notes: twiki

Previous MC Generator WG twiki: MonteCarloWorkingGroup

Full table of contents

Major updates:
-- JoshMcFayden - 2017-02-23 -- JoshMcFayden - 2017-03-10

%RESPONSIBLE% SimoneAmoroso
%REVIEW% Never reviewed

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2019-04-08 - SimoneAmoroso
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback