2.6. Central production of events

The central production of events for FCC physics studies is done using the EventProducer framework.

Get in touch with coordinators

In order to use the EventProducer, please get in contact with the FCC software and computing coordinators as running central production of events requires specific rights (see here).

When samples produced by the EventProducer framework are ready they will appear on this web-page.

2.6.1. Clone and initialisation

If you do not attempt to contribute to the repository, simply clone it:

git clone git@github.com:HEP-FCC/EventProducer.git

If you aim at contributing to the repository, you need to fork and then clone the forked repository:

git clone git@github.com:YOURGITUSERNAME/EventProducer.git

Then initialise:

source ./init.sh

In order to run batch generation, please add your CERN user name to the userlist in config/users.py

2.6.2. Generate LHE files from gripacks

To send jobs starting from a gridpack that does not exist but that you have produced, do the following:

  • place gridpack on eos

    • for FCC-hh /eos/experiment/fcc/hh/generation/gridpacks/

    • for FCC-ee /eos/experiment/fcc/ee/generation/gridpacks/

  • if the gridpack is from Madgraph, name it mg_process (and call option gp_mg when running generation commands), if from powheg please name it pw_process (and call option gp_pw),

  • add to config/param_FCCee/hh.py an entry corresponding to the gridpack name in the gridpacklist list, depending on the study.

If the gridpack already exists and has been properly added to the correct config/param_FCCee/hh.py, then simply run:

python bin/run.py --FCCee --LHE --send --condor --typelhe gp_mg -p <process> -n <nevents> -N <njobs> -q <queue> --prodtag <prodtag> --detector <detector>

example to send 10 jobs of 10 000 events of ZHH events at 365GeV using the longlunch queue of HTCondor for FCC–ee for the spring2021 production tag and the IDEA detector:

python bin/run.py --FCCee --LHE --send --condor --typelhe gp_mg -p mg_ee_zhh_ecm365 -n 10000 -N 10 -q longlunch --prodtag spring2021 --detector IDEA

The options --ncpus and --priority can also be specified to increase the numbers of cpus on the cluster and to change the priority queue.

2.6.3. Generate LHE files directly from MG5

To send jobs directly from MG5, you need a configuration file (see in mg5/examples directory *.mg5) and, optionally:

  • a cuts.f file (containing additional cuts)

  • a model (see in models directory for instance)

Nota Bene

At the moment no example is generated for FCC-ee this way. Below is an example for FCC-hh.

As before, you need to add the process to the config/param_FCChh.py file. Then you can run with the following command:

python bin/run.py --FCC --LHE --send --condor --typelhe mg -p mg_pp_hh_test --mg5card mg5/examples/pp_hh.mg5 --model mg5/models/loop_sm_hh.tar -N 2 -n 10000 -q workday  --memory 16000. --disk 8000.

The options --ncpus and --priority can also be specified to increase the numbers of cpus on the cluster and to change the priority queue.

2.6.4. Generate STDHEP files directly from Whizard + Pythia6

This section describe how to send Whizard + Pythia6 jobs and produce STDHEP files. The steps to be followed can be described as:

  1. Define process in gridpacklist in config/param_FCCee/.py.

  2. Write the Whizard sin card and put it in: /eos/experiment/fcc/ee/generation/FCC-config/<prodtag>/FCCee/Generator/Whizard/<version> by making a Pull Request to the corresponding production tag branch of FCC-config for example wzp6_ee_eeH_ecm240.sin

  3. Once the Pull Request is approved and the file has appeared on the corresponding <prodtag> on eos

  4. Send jobs

python bin/run.py --FCCee --STDHEP --send --typestdhep wzp6 --condor -p <process> -N <njobs> -n <nevents> --prodtag <prodtag> -q <queue>

Example produce 1 job of 10000 events of mumuH at FCC-ee 240GeV

python bin/run.py --FCCee --STDHEP --send --typestdhep wzp6 --condor -p wzp6_ee_mumuH_ecm240  -N 1 -n 10000 --prodtag spring2021 -q longlunch

2.6.5. Generate EDM4hep files from the LHE and decay with Pythia8

  1. If you want to let Pythia decay without specifying anything, you can use the default card, but if you have requested extra partons at matrix element, you might need to specify matching parameters to your pythia card

  2. If you want to use a specific decay, make sure that the decay you want is in decaylist and branching_ratios of the config/param_FCCee/hh.py

  3. Then create appropriate Pythia8 card, by appending standard card with decay syntax if needed and add it to the proper directory. For a given production tag of FCC-ee this directory is:

/eos/experiment/fcc/ee/generation/FCC-config/<prodtag>/FCCee/Generator/Pythia8/

Nota Bene

Please do not write directly on EOS. Cards should be added by making a documented pull request to the corresponding production tag branch in FCC-config.

  1. Send the jobs:

python bin/run.py --FCChh/FCCee --reco --send --type lhep8 --condor -p <process> -N <njobs> -q <queue> --prodtag <prodtag> --detector <detector>

Example produce 10 jobs of FCC Delphes events of ttz decaying the Z to neutrinos. :

python bin/run.py --FCCee --reco --send --type lhep8 --condor -p mg_ee_zhh_ecm365 -N 10 -q workday --prodtag spring2021 --detector IDEA

Please note that the decay in pythia is optional, and that there is no need to specify the number of events to run on as it will by default run over all the events present in the LHE file

The options --ncpus and --priority can also be specified to increase the numbers of cpus on the cluster and to change the priority queue.

2.6.6. Generate EDM4hep files from STDHEP

To produce Delphes EDM4hep files from STDHEP, just run this kind of command

python bin/run.py --FCCee --reco --send --type stdhep --condor -p <process> -N <njobs> -q <queue> --prodtag <prodtag> --detector <detector>

For example to run over one STDHEP file of mumuH process:

python bin/run.py --FCCee --reco --send --type stdhep --condor -p wzp6_ee_mumuH_ecm240 -N 1 -q longlunch --prodtag spring2021_training --detector IDEA

2.6.7. Generate EDM4hep files from Pythia8

This section describe the way most of the events were produced, using Pythia directly. The steps to be followed can be described as:

  1. Define process in pythialist in the config/param_FCCee/hh.py corresponding to your FCC choice.

  2. Write Pythia8 process card and put it in: /eos/experiment/fcc/ee/generation/FCC-config/<prodtag>/FCCee/Generator/Pythia8 by making a Pull Request to the corresponding production tag branch of FCC-config for example p8_ee_Zbb_ecm91.cmd

  3. Once the Pull Request is approved and the file has appeared on the corresponding <prodtag> on eos, send jobs

python bin/run.py --FCC-hh/FCCee --reco --send --type p8 --condor -p <process>  --pycard <pythia_card> -n <nevents> -N <njobs> -q <queue> --prodtag <prodtag> --detector <detector>

Example produce 1 job of 10000 events of ZH at FCC-ee 240GeV

python bin/run.py --FCCee --reco --send --type p8 -p p8_ee_ZH_ecm240 -n 10000 -N 1 --condor -q longlunch --prodtag spring2021 --detector IDEA

The options --ncpus and --priority can also be specified to increase the numbers of cpus on the cluster and to change the priority queue.

Important

If --pycard option is not specified, this step will run with the default Pythia8 card (in this case p8_ee_default.cmd), that does not include specific decays nor specific matching/merging parameters.

To assist you in writing your own Pythia8 configuration cards, the manual is available here

2.6.8. Expert mode

The following commands should be run with care, as they update the database, web-page etc… They run automatically every four hours with crontab, thus you will eventually know when your samples are ready to be used by browsing the corresponding configuration on this web-page. The --force option is used to force the script to run in order to optimize running time, processes that have not been flagged will not be checked.

2.6.8.1. Updating the database

  1. First one need to check the eos directories that have been populated with new files.

Example for LHE:

python bin/run.py --FCCee --LHE --checkeos [--process process] [--force]

Example for Delphes events:

python bin/run.py --FCCee --reco --checkeos --prodtag spring2021 [--process process] [--force]

  1. Second one need to check the quality of the files that have been produced. Example for LHE:

python bin/run.py --FCCee --LHE --check [--process process] [--force]

Example for Delphes events:

python bin/run.py --FCCee --reco --check --prodtag spring2021 [--process process] [--force]

  1. Then the checked files needs to be merged (individual yaml files are created for each file produced and the information needs to be aggregated): Example for LHE:

python bin/run.py --FCCee --LHE --merge [--process process] [--force]

Example for Delphes events:

python bin/run.py --FCCee --reco --merge --prodtag spring2021 [--process process] [--force]

2.6.8.2. Cleaning bad jobs

To clean jobs that are flagged as bad, the following command can be used for LHE:

python bin/run.py --FCCee --LHE --clean [--process process]

and for Delphes

python bin/run.py --FCCee --reco --clean --prodtag spring2021 [--process process]

As the code checks the files that are in the end written on eos, we need to clean also old jobs that don’t produced outputs 3 days after they started. To do so run the following command for LHE

python bin/run.py --FCCee --LHE --cleanold [--process process]

and for Delphes

python bin/run.py --FCCee --reco --cleanold --prodtag spring2021 [--process process]

If you want to completly remove a process, the following command can be used with care for LHE:

python bin/run.py --FCCee --LHE --remove --process process

and for Delphes

python bin/run.py --FCCee --reco --remove --process process --prodtag spring2021

2.6.8.3. Update the webpage

The webpage can be updated after the files have been checked and merged by running for LHE

python bin/run.py --FCCee --LHE --web

and for Delphes

python bin/run.py --FCCee --reco --web --prodtag spring2021

2.6.8.4. Create the sample list for analyses

To create the list of samples to be used in physics analyses for spring2021 production tag:

python bin/run.py --FCCee --reco --sample --prodtag spring2021

2.6.8.5. All in One

a script can be used to do all the steps in one go, for example for LHE at FCC-ee:

./scripts/cronrun_LHE_FCCee.sh

and for Delphes events:

./scripts/cronrun_RECO_FCCee.sh <prodtag> <detector>

for example for Delphes events with spring2021 production tag and IDEA detector:

./scripts/cronrun_RECO_FCCee.sh spring2021 IDEA

2.6.9. Produce samples with EventProducer outside of official campaign

This section explains how you can use the EventProducer to produce your own database of events.

  1. You first need to create three output directories with sufficient disk space. One where the samples could be written, for example on your CERN eos. An other public directory to store the database that is created and associate each job/file on eos, and one that can be browsed from the web, for example using your eos cern box for all:

For samples  /eos/home-<X>/<cernlogin>/generation/
For public   /eos/home-<X>/<cernlogin>/public/FCCDicts/
For web      /eos/home-<X>/<cernlogin>/www/data/FCCee/

  1. Edit the file config/param_FCCee.py:

    • Replace webbasedir by your directory For web

    • Replace pubbasedir by your directory For public

    • Replace eosbaseoutputdir by For samples

    • If you want to use your own FCC-config, also replace eosbaseinputdir to point to where it is cloned.

  2. You should now be ready to send jobs using your own work areas, you will have to run the checks yourself, have a look at the Expert mode