OSG Rho Simulations

From GlueXWiki
Revision as of 15:07, 29 April 2021 by Beattite (Talk | contribs) (Other Information)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Information for Richard Jones regarding generation of rho MC using MCWrapper on the OSG.


Uploading these files automatically capitalized the first letter. The first letter of file 2 (MC_rho.config) and file 5 (Gcontrol_HADR4.in) are meant to be capitalized, but the remaining four should not be when used in the provided script. The script is still set up with the paths I used on the ifarm, so they should be changed to match your system.

  1. Media: Gen_2pi_flat.cfg
  2. Media: MC_rho.config
  3. Media: Jana_analysis_2017-01_ver38.config
  4. Media: Recon-2017_01-ver03_26.xml
  5. Media: Gcontrol_HADR4.in
  6. Media: Analysis-2017_01-ver38.xml

Alex A.'s Original Email

"You will find a gen_amp config file (gen_2pi_flat.cfg), which will produce a BW-shaped rho with an isotropic decay into pi+pi- (i.e. flat). The first line of the config file supplies some command line options to MCWrapper. Here I select the beam energy range (8.2-8.8) and put the minimal t to 0.05 GeV2 because we don't have acceptance for exclusive reactions below that.

The second file (jana_analysis_2017-01_ver38.config) is the jana config file for the analysis step. It has broadened timing cuts and I just removed all other reactions except pi+pi-.

For the main generation and reconstruction part (ENVIRONMENT_FILE in MC.config), you should use the latest edition of the one corresponding to the REST version, currently recon-2017_01-ver03_26.xml. For the analysis part (ANA_ENVIRONMENT_FILE), you have to supply the correct xml file which was used for the launch: analysis-2017_01-ver38.xml"

Other Information

In testing, running MCWrapper multithreaded resulted in an hd_root error (Exit code: 70) for which I couldn't find the meaning or solution. To avoid this, we can run single-threaded. I made the appropriate change to file 3 (the jana configuration file) to run hd_root with one thread.

The command I ran on the ifarm in testing was:

  • /group/halld/Software/builds/Linux_CentOS7.7-x86_64-gcc4.8.5/gluex_MCwrapper/gluex_MCwrapper-v2.5.1/gluex_MC.py MC_rho.config 30730-30788 2000 cleanmcsmear=0 numthreads=1

This generated 2000 events using MC_rho.config (file 2) with run numbers ranging from 30730 through 30788, which corresponds to the run numbers available from the 2017 ver38 analysis launch. It saves the mcsmear output and runs with one thread. You can also use the per_file=N option to set the number of events to be stored per hddm file if you'd like (the default is 10000, I use 25000 or 50000 if I'm doing larger batches of particle gun data).

The output files of interest are the smeared HDDM files and the REST trees.

We wish to generate the same amount of data with HDGeant4 and also with HDGeant3 HADR4. Changing between the two can be done with the GEANT_VERSION flag in MC_rho.config.

As a reminder, we also want to look at the TrajectoryPoint output, so the MC generation should be run with the custom store_trajectory_ancestry branches of HDGeant4 and HDGEANT3 (I did not alter the provided xml files to reflect this).

I think something like 400k events is the minimum we'd want to generate, but if generation on the OSG is fairly fast and you can do something like 1M, that would be nice.