sherpa's status and plans in cms

19
The CMS SHERPA Team Gurpreet Singh Chahal (IPPP Durham University and Imperial College-UK) Jie Xiao and Yang Heng(IHEP-China & DESY) Si Hyun Jeon and Hyon San Seo ( Seoul National University-South Korea) Siew Yan Hoh (Padova University-Italy) Wajid Ali khan (NCP-Pakistan) Amandeep Kaur (Panjab University-India) SHERPA's Status and Plans in CMS The Annual SHERPA Meeting, January 2020 https://indico.cern.ch/event/835664/timetable/

Upload: others

Post on 18-Dec-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

The CMS SHERPA Team • Gurpreet Singh Chahal (IPPP Durham University and Imperial College-UK)

• Jie Xiao and Yang Heng (IHEP-China & DESY)

• Si Hyun Jeon and Hyon San Seo ( Seoul National University-South Korea)

• Siew Yan Hoh (Padova University-Italy)

• Wajid Ali khan (NCP-Pakistan)

• Amandeep Kaur (Panjab University-India)

SHERPA's Status and Plans in CMS

The Annual SHERPA Meeting, January 2020

https://indico.cern.ch/event/835664/timetable/

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Sherpa is included in CMSSW as external package: https://github.com/cms-externals/sherpa

• Uses a couple of additional external packages:

• HepMC, LHAPDF, Sqlite, FastJet, ... (used by many others)

• BlackHat, Openloops, ... (used only by Sherpa)

• Releases • Current recommended versions: SHERPA 2.2.8 + OpenLoops 2.1.0 • These versions are available in various CMS software versions

• SHERPA samples can be produced for upgrade studies and all the data taking periods in Run-2,

i.e. 2016, 2017 and 2018

• SHERPA is part of CMS generator validation & production infrastructure

• ZtoEE NLO and ttbar NLO + 2 LO jets

• W+9jets is an interesting case to be added to validation workflow to further understand its

physics & computing performance

2

SHERPA in CMS - I

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 3

| Philipp Millet | Sherpa Tutorial | 29.08.2016 |

| Using Sherpa in CMSSW |

THE CMSSW SHERPA INTERFACE – OVERVIEW

5

• the Sherpa Interface is part of CMSSW

� source code:

https://github.com/cms-sw/cmssw/tree/CMSSW_8_1_X/GeneratorInterface/SherpaInterface

• purpose:

� generate events with Sherpa

� turn events into compliant format

� feed output into CMSSW

• Sherpa-CMSSW workflow

� phase-space integration and library creation

� event generation

� performed by interface

� can be parallelized on grid

details next slides

details later

SHERPA in CMS - II

https://github.com/cms-sw/cmssw/tree/CMSSW_9_4_X/GeneratorInterface/SherpaInterface

• Documentation: HEPForge is hosting the releases until 2.2.5 but SHERPA is moved to Gitlab for the future releases • SHERPA workshop: https://indico.cern.ch/event/779232/ • CMS Twiki: https://twiki.cern.ch/twiki/bin/viewauth/CMS/SherpaInterface • S u p p o r t : P r i m a r i l y r e p o r t i s s u e s a t G E N h y p e r n e w s : h n - c m s - g e n e r a t o r s @ c e r n . c h

or at Sherpa group: [email protected]

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Samples getting ready to be submitted in central production

• DYJetsToLL Pt-*to* (Pt binning)

• WJetsToLNu (jet binning)

• For MC production of private and central samples

• Plan to post the run cards on the official GitHub SHERPA-GEN area: https://github.com/cms-sw/

genproductions/tree/master/bin/Sherpa/cards/production

4

SHERPA Samples

• SHERPA samples are being produced for the 17+ analyses

• For analyses in standard model, top physics and Higgs physics and Exotica groups

• Spreadsheet to document the Sherpa sample's usage in CMS (private and central)

• About 1.3 billion+ events are expected to be sent in production

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• The First CMS SHERPA workshop was organised

• Almost monthly SHERPA meetings to understand the studies about physics and computing

performances of various samples

• Major interest received from precision measurement groups (standard model physics, top

physics and Higgs groups)

• A few studies performed for beyond standard model analyses as well

• B-physics group is interested to do the comparison studies of Sherpa vs other generators

5

SHERPA activities in 2019

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 6

Physics and computing performance comparison of leading generators

• Following the Physics Event Generator Computing Workshop, the work started to get a fair comparison

of Sherpa vs madgraph vs other leading generators to understand the computing performance.

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 7

Physics and computing performance comparison of leading generators

• CMS plans to present the more mature comparison studies in the coming HSF meetings this year

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Preliminary studies to understand the physics

and computing performances

8

ttV(ll/lv/vv/qq) + 0,1 jets @ NLO + 2 jets @ LO

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Sherpa v2.2.8 with OpenLoops v2.1.0

• Compare gg fusion higgs process with different Sherpa

version (v2.2.6)

• Compare gg fusion higgs process to other generator

(MadGraph)

• Event generation and gridpack production time

9

ggF Higgs + LO012/NLO01 jets

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 10

ttbar + 0,1jet @ NLO & 2,3,4 jets@ LO

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• ZZ4l samples for NTGC model are being studied

• Signal processes for Dark photon model

11

Samples for Exotica physics

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• ttbar + 0,1 jet at NLO, 2,3,4 jets at LO

• Single top processes

• WW+0(,1) jets @ NLO + (1,)2(,3) jets @LO

• Sherpa+FastNLO for W+Jets αs(MZ)

• V+jets with NLO EWK corrections

• NLO photon + jets

• Z+jets for Inclusive jets cross-sections for pp collisions at 13TeV

• Higgs+jets production in gluon fusion with finite top mass effects

• VBF process to study the impact of parton shower uncertainties and compare different Parton

showers for Higgs XSection working group

• ZZ4l aTGC production, enabling the production of b enriched samples, etc.

• B physics processes

12

Interests and ongoing work - I

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• VBS: SMEFTSim UFO model studies with Sherpa

• Planning to organising a Sherpa workshop and Sherpa tutorials

• Two MCnet positions are being filled to study:

• Studying systematic uncertainties for Higgs process with Sherpa

• Physics and computing comparison studies for B physics process with Sherpa and other

leading generators

• Rewriting/updating the Sherpainterface to CMSSW to improve its computing efficiency and

making it more user friendly

• Developing CMS specific tunes for Sherpa

13

Interests and ongoing work - II

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Recent versions of SHERPA, Openloops are integrated in ~1 month timescale in

CMS software

• SHERPA samples can be produced for upgrades or any data taking period in Run-II

• CMS is interested to invest more manpower to profit from Sherpa

• Sherpa interface to CMSSW will be updated

• To get ready for Run-III analyses

14

Summary

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 15

Back Up

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 16

Overview

User Inputs Parton ShowersMatrix Elements Soft Physics

The SHERPA 2.2 event generator framework

Initial Beams

collider setupPDFs (built-in,LHAPDF)          beam spectra

Physics Process

parton levelperturbative order(QCD/EW)selectorsmatching/mergingpartonic decays

Parameters/Models

FeynRules/UFOcouplingsmassesvariationsshower settingsnon-perturbativeparameters

Matching and Merging

Output Formats

HepMC              LHEFRoot Ntuple

Interfaces

RIVET analyses   C++/Python ME accessMCgridintegration intoATLAS/CMS

Code/Docu

HepForgeGitLabonline documentation 

         sherpa.hepforge.org

gitlab.com/sherpa-team/sherpa

CS-Shower(default)

dipole showerfully massiveQED splittings

 DIRE

hybrid dipole-partonshower algorithmfully massive

Matrix Element Generators

AMEGICCOMIXCS subtraction

1-loop Amplitudes

OpenLoopsRecolaGoSamBLHA

     Automated MC@NLO style matching

     Multijet-merging algorithms

based on truncated showerstree-level and one-loop matrix elements:  MEPS@LOand MEPS@NLO  approximate electroweak corrections

     NNLO QCD with parton showers

selected processes only

Hadronisation

AHADIC: a clusterfragmentation modelinterface to Pythiastring fragmentation

Hadron Decays

decay tables forhadronic resonancesdedicated form-factormodels, e.g. τ, B, Λspin correlationsYFS QED correctionspartonic channels

Underlying Event

multiple partoninteractionsbeam-remnant coloursintrinsic transversemomentum

Interfaces/Outputs

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 17

Overview Recent results Usage Running SHERPA Conclusions

The SHERPA event generator framework• Two multi-purpose Matrix Element (ME) generatorsAMEGIC++ JHEP02(2002)044, EPJC53(2008)501

COMIX JHEP12(2008)039, PRL109(2012)042001

• Two Parton Shower (PS) generatorsCSSHOWER JHEP03(2008)038

DIRE EPJC75(2015)461

• A multiple interaction simulationa la PYTHIA AMISIC++ hep-ph/0601012

• A cluster fragmentation moduleAHADIC++ EPJC36(2004)381

• A hadron and ⌧ decay package HADRONS++

• A higher order QED generator usingYFS-resummation PHOTONS++ JHEP12(2008)018

Sherpa’s traditional strength is the perturbative part of the eventLO, NLO, NNLO, LOPS, NLOPS, NNLOPS, MEPS, MENLOPS, MEPS@NLO

Marek Schonherr SHERPA-2.2.5: overview, developments and usage 4/36

JHEP02(2009)007

The SHERPA Event Generator Framework

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020 18

Overview Recent results Usage Running SHERPA Conclusions

Acronyms and nomenclatureFixed order calculations

• matrix elements only, implies fixed multiplicities

• no parton shower, no non-perturbative physics, no particle level

) LO, NLO, NNLO

Parton shower matched calculations

• combination of fixed order calculation and parton shower for onemultiplicity

• particle level predictions, no multijet observables

) LOPS, NLOPS, NNLOPS

Multijet merged calculations

• combination of parton shower matched calculations for increasingfinal state multiplicities (mostly jets)

• particle level predictions, multijet observables

) MEPS(@LO), MEPS@NLO (special case MENLOPS)

Marek Schonherr SHERPA-2.2.5: overview, developments and usage 5/36

Acronyms and Nomenclature

Gurpreet Singh ChahalThe Annual SHERPA meeting, Milano 8 Jan 2020

• Physics Event Generator Computing Workshop on 26-29 November. • To get a fair comparison of Sherpa vs madgraph vs other leading generators to understand the

computing performance. • The CMS Sherpa contacts have already started it. Anyone is welcome to join the effort!

• Frank Siegert - ATLAS contact • Gurpreet Singh Chahal - CMS contact • Olivier Mattelaer - MG5 contact • Stefan Hoeche - Sherpa contact

• We will cross check the cards, etc. to help to get these studies done. • We should have a uniform timing assessment that could be obtained from running the generation on

private machines. The initial list of processes is here that can be extended later: • Z/W + jets, LO merged with up to 4 jets and (separately) NLO/LO merged with up to 2 / 4 jets

(Siewyan, Sihyun, Jie and Cheng) • tt~ + jets, LO merged with up to 3 jets and (separately) NLO/LO merged with up to 2 / 3 jets (Wajid)

• Produce 100 k event for the LO processes or less for more complex processes. • Add these two additional flags in Sherpa run cards:

• PP_RS_SCALE VAR{sqr(sqrt(H_T2)-PPerp(p[2])-PPerp(p[3])+MPerp(p[2]+p[3]))/4}; • PP_HPSMODE 4;

• The summary of initial plan is available in this google doc

19

Performance Comparison of Generators