Skip to content

Latest commit

 

History

History
133 lines (103 loc) · 8.09 KB

quickplot_instructions.md

File metadata and controls

133 lines (103 loc) · 8.09 KB
Raw

Instruction to produce a new beamspot payload

These instruction explains how to fit a new beamspot in a given run, alignment and release, and how to create a new payload from the fit results for a new global tag. An example is described at the end.

work space setup

Download your favourite CMSSW release, e.g. 13_2_0_pre3

cmsrel CMSSW_13_2_0_pre3
cd CMSSW_13_2_0_pre3/src
cmsenv

Fetch the beamspot tools and compile

git-cms-addpkg RecoVertex/BeamSpotProducer
cd RecoVertex/BeamSpotProducer/python
git clone https://github.com/MilanoBicocca-pix/BeamspotTools.git
cd $CMSSW_BASE/src
scram b -r -j8

The uncertainty python module is used in some scripts. It should be included in recent CMSSW releases, if not:

pip install --user uncertainties

beamspot fit

The beamspot fit is configured by the BeamFit_custom_workflow.py file, which is an extension of the standard BeamFit_*_Workflow.py configuration files. To run the fit:

cmsRun BeamFit_custom_workflow.py     \
  jobName="name"                        \
  globalTag="Global Tag name"                       \
  refit=bool                            \
  runs="run:ls-run:ls"                  \
  dataset="/dataset/period/tier"        \
  storage="root://service"              \
  highPurity=bool                       \
  tracks="collection_name"              \
  updateGT="record1:tag1:label1,record2:tag2:label2" \
  localPL="sqlite_file:/path/to/local/file" \
  saveRootFiles=bool

where:

  • jobName is the string used to create the output file names (default: beamspotFit).
  • globalTag(*) is the global tag (default: '').
  • refit decides whether to re-run the track refit and consequently the vertex finding (default: false). Remember to set true when changing the tracker alignment and whatnot.
  • runs is the run range in the form "run:ls-run:ls,run:ls[...]" (default: 0:min-999999:max - all runs).
  • dataset(**) is the input dataset. The code will fetch the input files from DAS. If the inputFiles argument is passed, it will be used instead (default: ''). The /StreamExpress/Run*/ALCARECO and /ZeroBias/Run*/RAW* datasets can be used.
  • storePrepend in case files should be fetched from a specific store (default: '' - auto)
  • highPurity decides whether to use a high purity track selection (default: false)
  • tracks is the input track collection, which depends on the type of input dataset (default: generalTracks)
  • updateGT is used to update records, tags and labels in the global tag. Defined in the form "record1:tag1:label1,record2:tag2:label2,[...]".
  • localPL is used to specify a local payload for updating the GT. NOTE works only when --updateGT is set.
  • saveRootFile decides whether to save the fit results to a .root file. Usually needed only for debugging.

(*)   the argument is mandatory
(**) the argument or --inputFiles is mandatory

The result of the fit will be a .txt file containing the list of runs and the fit results in some format not compatible with the CMS database (see below how to convert this file to a .db file).
If the saveRootFile argument is set to true, a .root file will be created containing the fit results. Both the .txt and .root file name are set by the jobName argument.
A file with the "_filelist.txt" prefix is also creted. It contains the list of files used in the fit (NOTE: if specific lumisections are analyzed by setting the runs argument, the fitter might consume more files than needed as there is no way to fetch single lumisections from DAS. These files will also appear in this list but their content will be skipped).
(NOT YET) A file with the "_cfg.py" prefix will be written and will contain the full configuration of the job.

NOTE the track collection from ALCARECO is ALCARECOTkAlMinBias, for RAW the label is generalTracks.
NOTE Other parameters of the fitter are present in the configuration file and usually there is no need to modify them.
NOTE by default, the fit is run on one lumisection. This can be changed by editing the parameters (note: set them equal).
NOTE if a track refit is done and the alignment is not changed, the beamspot result will still be different. This is because, in such case, the configuration introduces additional cuts on the PVs. NOTE the fitter appends the results of the fit to the .txt file live, thread safety is NOT guaranteed.

beamspot plot

NOTE the plotting script is a work in progress. At the moment it can plot a single file in a format good for check-ups (bs parameter vs. time). NOTE this script does not work with any CMSSW release. Python 3 is required. All the modules used should be available in the default python3 distribution, except ROOT.
NOTE auth-get-sso-cookie is required. This is available by default on lxplus or can be installed user-side following the instructions of the repo.
NOTE a valid kerberos ticket is required to download data from OMS.

To plot the results, run plotFromTxt.py:

python3 plotFromTxt.py --input input_file_list --output output_dir --streams 5 [--canfail]

where

  • --input is the list of input .txt files produced by the BeamSpotProducer (NOTE currently works with a single file)
  • --output is the output directory
  • --streams is the number of parallel requests made to OMS (default: 1)
  • --canfail is a boolean which allows to select failed BS fits (type!=2 in the .txt file) (default: false)

The script will communicate with OMS to fetch date information for each lumisection and other paramters of the fill. This may take some time, depending on the number of lumisections selected.
The script will create the output directory and plot the beam spot parameters as a function of time (run / lumisection) in epoch format.

payload creation

There are three records where the BS information are stored

  • BeamSpotObjectRcd used offline
  • BeamSpotOnlineHLTObjectsRcd used online
  • BeamSpotOnlineLegacyObjectsRcd used online All the three records should be updated before a FTV.

Then, to create pb files:

  • OFFLINE: To convert the .txt file to a .db file use write2DB.py following the instructions here. The offline record is BeamSpotObjectRcd.
  • ONLINE: To convert the .txt file to a .db file use BeamSpotOnlineRecordsWriter_cfg.py. The online records to be updated are BeamSpotOnlineHLTObjectsRcd AND BeamSpotOnlineLegacyObjectsRcd

example

We want to fit the beamspot on ALCARECO files for run 370772 of 2023D, updating the tracker alignemtn tag to TrackerAlignment_collisions23_forHLT_v5. We will use a single local file from the /StreamExpress/Run2023D-TkAlMinBias-Express-v2/ALCARECO dataset (to run on the full run, use --dataset=dataset_name instead of --inputFiles)

xrdcp root://cms-xrd-global.cern.ch//store/express/Run2023D/StreamExpress/ALCARECO/TkAlMinBias-Express-v2/000/370/772/00001/63cbf285-ce54-4ad2-bc9b-833799330067.root AlcaFile_2023D.root
cmsRun BeamFit_custom_workflow.py         \
  jobName="localTest_refit_alignmentUpdate" \
  inputFiles="file:AlcaFile_2023D.root"     \
  runs="370772:min-370772:max"              \
  globalTag="130X_dataRun3_Express_v3"      \
  tracks=ALCARECOTkAlMinBias                \
  refit=True                                \
  updateGT="TrackerAlignmentRcd:TrackerAlignment_collisions23_forHLT_v5:"

The result can be compared to running without the --updateGT argument to see the effect of the new alignment.

To plot and check the result:

python3 plotFromTxt.py --input localTest_refit_alignmentUpdate.txt --output Run2023_370772 --streams 10