[Hps-analysis] FW: MC production

[Nathan Baltzell]

On this point:

> We need to decide where to generate. Since wab-beam-tri requires a large disk space, JLab might be the only possibility. 

If the size of the generator files is an issue at SLAC, do they really need to be kept during MC production at SLAC?  Is saving them just a convenience / time issue for when rerunning simulation with a different detector / readout / recon?  (I thought the large majority of cpu time was spent on readout anyway?)



On Jun 13, 2017, at 16:18, Maruyama, Takashi <tvm at slac.stanford.edu> wrote:

> Any comments are welcome.
> 
>  Takashi
> 
> -----Original Message-----
> From: Maruyama, Takashi 
> Sent: Tuesday, June 13, 2017 10:26 AM
> To: Bradley T Yale (btu29 at wildcats.unh.edu); McCormick, Jeremy I.; Graf, Norman A.
> Subject: MC production
> 
> Thanks to Jeremy, MC production is getting ready for mass production at SLAC.  Since many changes have been made recently, I would like to clarify MC parameters and make proposal for mass production. I would like to present this at the analysis meeting or HPS weekly meeting in the next few weeks.  Any comments/additions/corrections are all welcome.
> 
> 1) Generator level cuts
>      EGS5 (beam_v5.f)
> 	- FEE cut uses 15 mrad. Do we need to lower to 5mrad?
>                - wab cut uses 10 mrad. Do we need to lower to 5mrad?
>      MadGraph
> 	-Tritrig
> 	      alpha=1/137.036
> 	      E(e+,e-) > 50 MeV
>                      E(e+)+E(e-) > 0.5 GeV
>                      |theta_y| > 5 mrad
>                -Wab
>                      alpha=1/137.036
>                      E(gamma) > 50 MeV, |theta_y(gamma)| > 5 mrad
>                -ap
> 	     No cuts, but we need to specify A' mass and ctau.
> 
>      Generator level event generation can be started at SLAC, but we have to make sure the initial random number seeds are different between JLab and SLAC so that we don't duplicate events.
> 
> 2) Event generation
>        1.05 GeV
>             Beam background: 100 jobs, 500K bunches each
>             Tritrig: 1000 jobs, 10,000 events each
>             Wab:  1000 jobs, 10,000 events each
>             Rad: 100 jobs, 10,000 events each
>             Ap: 100 jobs, 10,000 events each for each mass.
> 
>        2.3 GeV
>              Same as 1.05 GeV
> 
> 3) Detector models
>      There are ~100 detector models. We need to decide which detector model to use for mass production.
>      Please make the list.
> 
> 4) Which files to save and where.
>     All the generator level files are saved, but only recon files are saved.
>     Since tapes cannot be used at SLAC, do we want to transfer all the recon files to JLab?
>     At SLAC, the files will be saved in /nfs/slac/g/hps3/data/mc_production . This directory tree may need a clean-up.
> 
> 5) What to generate and at what priority.
>       1.05 GeV
>               Wab-beam-tri using nominal detector (priority #1)
>               Tritrig-wab-beam using nominal detector (priority #2)
> 
>       2.3 GeV
>               Wab-beam-tri using nominal detector (priority #3)
>               Tritrig-wab-beam using nominal detector (priority #4)
> 
>       4.4 GeV
>               Wab-beam-tri using L0 detector
>               Tritrig-wab-beam using L0 detector
> 
>       We need to decide where to generate. Since wab-beam-tri requires a large disk space, JLab might be the only possibility. 
> 
> 6) Spreadsheet
>      We need a spreadsheet based on Google doc that summarizes all the files and production status/priorities.  This spreadsheet must be accessible by all the HPS collaborators (read only).
> 
> 
> Takashi
> 
> 
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis