From now on production meetings have a room booking:¶
- Mondays - Nu's Room starting Sept. 22 from 11:30a - 1:00p - reservation ends on Dec. 22, 2014 * Thursday - Theory Room starting Sept. 25 from 11:30a - 1:00p reservation ends on Nov. 20, 2014 3rd floor NW (Please note that this room will not be available on Oct. 23 due to the NOvA Collaboration Meeting.)
Matthew, Jonathan, Chris, Nitin, Pavan, Jeny, Michael, Nate, Gavin, Kirk, Ryan, Craig, Adam
It was discussed in the conveners meeting that stream merging could proceed in the raw2root step and to facilitate this Mark Messier has a student who can work on it. How to do this technically is less clear, would we have one dataset per run? Or use a custom script? Input files will need to be ordered. It also became clear that meta-data merging won't work out of the box for any nova specific meta-data, e.g. online.XYZ.
Action Item: Matthew will follow up when at FNAL next week.
The keep up sample sample copier is running as NoVA pro on gpsn01:
00 * * * * source /nova/app/home/novapro/cron/daily_cron.sh &> /nova/app/home/novapro/cron/cronlog_daily_cron.txt
With output datasets written to:
This will be reviewed tomorrow, then if all is well switched to run daily and announced to the collaboration. Finally the list of principles for ProductionScripts was circulated and everyone will respond to confirm what is there. These can then be added to ProductionScripts and the switch over can happen.
Towards a simulation & reconstruction tag and plans for submissions (Matthew, Nate)¶
There are some outstanding questions about whether genie reweighting is working in the current release. The plan was to fix this later in the day and discuss during the simulation meeting. As of Tuesday morning all seems well and everyone has signed off on the creation of a new tag.
Action Item: Jonathan will make a new tag off of the first-ana branch "FA14-09-23".
Action Item: Matthew will run the production tests as soon as this tag is ready.
Action Item: Nate will make the Simulation validation files. When these are done we will launch the big production!
Continued discussion on systematic (geo-jittered) samples (Ryan et al.)¶
It seems inevitable that we will want the machinery to reuse events for systematic studies at some point, so a plan was formulated to address this. Using genie with fixed random seeds at the start of a job will not produce identical events with different geometries as the change in geometry will lead to a divergence of these and hence the same events will not be reproduced. The idea instead is to run a job which strips the GEANT+ stages out of a SIM file (and could in future strip out just photon transport for instance) then re-runs GEANT with an altered geometry re-using the underlying genie events. The hope being that the milli-meter scale differences in where interactions happen would end up below our observable resolutions.
Action Item: Nate will study this later in the week once the first analysis simulation jobs are up and running.
Mr-Brem update (Duyang, Nitin)¶
Nitin reported that they're witnessing 30% failure rates in local jobs. He will work with Gavin to reduce this before we're ready to proceed.
Additional requests for keep-up reconstruction (Matthew et al.)¶
Kirk has committed most of what is needed for cosmic rejection and nu_mu. He's also altered these so that they can run without the need for PID. It was discussed that the option to fall back from a PID-ed track to the longest one should be kept behind a FHiCL switch. Kirk will implement this. Dominick has also made some changes, such as adding cosmic veto and telling it to not obey preselection, and also adding the analysis sandboxes to the CAFs (but dropping them from the RECO files for easy retune-ability).
Keep-up update: PC-hits (Satish, Paola)¶
- Paola successfully ran some tests and executed a project, however this seems to take ~1.5 hours per file when it should take 30 minutes. She if following this up. * Remove beam spill has been added in by Satish, tested and looks good. * The datasets Satish defined for pchits and pchits stoppers are reversed, this problem goes back a couple of months. He will delete the one definitions and make new ones.
Keep-up update: Reco (Jeny)¶
Jeny has reviewed all the scripts and is happy with them, she also ran a test and Gavin reviewed the output. She is ready to proceed when we are.