[Hps-analysis] MC Trident Discrepancy

[Bradley T Yale]

Hey everyone,

This could be a major cause of the discrepancy in relative trident yields in MC.


The way I've been reading out the pure MC components is 100 generated files -> 1 recon file, but it seems that this was too many events for the readout step to handle, and it only processes a certain number of files before stopping. However, the jobs never report an error and complete fine while using <100 files. This will greatly affect the number of events used for normalization.


If looking at Pass2 MC (tagged '3.4.0'), then the number of generated files each recon file corresponds to is:


BH: 1 recon file = 63-100 generated files (depending on whether the fieldmap was applied)

RAD: 1 recon file = 16-30 generated files

tritrig: 1 recon file = 10 generated files


This can be easily fixed by standardizing the readout of everything without background to 10-to-1, which I will start doing in the pass3 folder, and that number of files can then be assumed.

In the meantime, the number of input files and events used for pass2 MC readout can be found listed in the corresponding logs and your calculations adjusted accordingly:

/work/hallb/hps/mc_production/pass2/logs/readout/


I am very sorry for not catching this sooner.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/hps-analysis/attachments/20151016/899e41d8/attachment.html>