Project

General

Profile

08/18/2014

Notes taken by M.Tamsett

Attendance

Matthew, Gavin, Andrew, Michael, Nate, Ryan, Chris, Craig, Jonathan, Paula, Jim, Dominick, Kanika & Satish.

Simulation submission planning

This meeting was mostly about educating Matthew. The first stage of simulation is the creation of FHiCL dataset. This is done by an individual running make_sim_fcl interactively to create a large number of FHiCL files and using the options in make_sim_fcl to send these to the hex drop boxes for SAM declaration.

Action Item: Matthew, collect and document the series of make_sim_fcl commands that are needed to create these datasets.

In the past Eric handled the submission of all MC simulation steps that run off of the aforementioned FHiCLs. Nate has kindly volunteered to do this this time around, although he will be away until September 4th. Gavin has agreed to act as his back-up if we start on the 1st. In the future we could move to a system where we make the FHiCLs then request SCD to run the simulation jobs. But not this time round. The required POT weighting options have been added in by Ryan, but we're still waiting on Joao to finalise the good runs list before this is ready to go. The plan for the first submission was discussed. Do we just submit jobs with the data-sample to shutdown, or do also we take an educated guess at what the potential top-up dataset data could look like?

Action Item: Matthew, meet with Eric and confirm which scripts he uses to submit jobs. Possible candidate is this.

The idea is to run a production at the same scale as February production. The RHC sample should be run as part of the main production as although this won't influence the go into prediction it will be used to make other plots.

SIM validation is currently executed using Gareth's scripts. Nate will work closely with him to validate this round.

Supporting samples

ND beam MC with low beam intensity

This will be needed later on, so we don't need to worry about it for the first submission. High intensity samples will likely not be needed. There was some discussion of the PoT per pulse that will be used to simulate the base-line ND beam MC. Apparently we're running 5e13 protons-per-pulse now, and the likely value will be 2.5e13-ish. Nate is in discussion with beam experts.

Action Item: Nate continue to follow up on PPP for first analysis.

Different FD geometry

The GDML FD beam MC with different "alignment" in geometry is available now (although support for it needs adding into make sim fcl). The plan as of 19/08 is to use FHiCL switches to run reco with the a different geometry down the line.

Action Item: Ryan add support for this into make_sim_fcl.
Action Item: Nate submit jobs for these.

GENIE+CRY simultaneous generation

These files look good.

Action Item: Ryan add support for this into make_sim_fcl.
Action Item: Nate submit jobs for these.

MRE/MRCC/Cosmic Brem

The need for these will be discussed at the conveners meeting with the analysis groups. Note this has never done as production before. Kanika is currently testing the steps needed for us to run these jobs.

FD rock

It was decided to assume that we won't need these. Ryan will follow this up.

Gavin’s slides.

Gavin was attempting to fix the problems with the ND CAF sample (light levels and overlay repetition). As part o this efforts he spotted (and fixed some) a number of problems with our current methods. Including more-metadata woes. It was noted that the occupied spill filter should not be used. Metadata is blocking FTS declaration due to art issues. He is trying to work out how long it will take to remake this sample.

Action Item: Gavin will follow up with Adam how he hacked his way around the meta-data issues (although Adam is on holiday)
Action Item: Gavin will reprocess these files in the original releases with just the fixes implemented, e.t.a of sample ~2 days (this is a 20% size sample I think).
Action Item: Fix the other issues before the sim tag stage.

AOB

Dominick says that a new version of libwda will be rolled out on Friday - this should fix memory allocation problems. He will follow this up.