Documentation of CLAS12 RICH simulation.
If working on ifarm (recommended), gemc/5.12 now includes all recent changes to the RICH geometry and digitization, and can be loaded like all other modules, module load gemc/5.12. The most recent additions by Marco to the ccdb should also mean that the online ccdb loaded will be set up correctly for the RICH.
Running gemc locally/on ifarm then requires
- clas12-config to load the correct gcards and coatjava yaml files
- clas12Tags to load the already-built CLAS12 geometry (tag 5.12)
In my setup on Duke's cluster, I would first unpack/clone clas12Tags, then clone clas12-config within the clas12Tags directory for ease of use with the relative paths as they are set in the gemc gcards.
Running gemc and coatjava should now be pretty straightforward (previously the RICH required a lot more checking of geometry and configuring the ccdb). On ifarm, you can use the already built gemc/5.12 executable. The example slurm job submission scripts in this directory will give an example of running gemc with a given lund file, and show exactly how I generated the most recent large simulated dataset used for calibration.
If you want to filter for only a specific set of needed output banks (recommended), you can follow the instructions in the coatjava yaml files (setting up a directory with links to the .json bank definition files).
I placed my simple python script for skimming lund files for the scattered electron (always kept for timing) and hadrons in sectors 4/1 in the lundSkim/ directory. It also includes an option for keeping events with the scattered electron in the RICH.
To run clasdis on ifarm, first module load mcgen. Then, clasdis can be run like, for example,
clasdis --docker --trig 10000 --zpos -3,
which will produce a file clasdis.dat with the 10000 lund-formatted events, and the vertex z position at -3cm. The directory lund-skim in this repository demonstrates how I have been producing these files with clasdis and skimming them for selected events.
- The slurm script I placed here for running gemc and coatjava is configured for Duke's cluster (should only require changing account and partition names)
- If anyone wants instructions for running in a container on a non-ifarm cluster, I am happy to provide it. I tried to write this up for use on ifarm since it is the most straightforward.