Architecture Review Meeting

Wednesday 1 Jul 2015, 15:00 → 16:00 US/Central

WH7XW (Woodshed) (Fermilab)

Audio connection will be opened with ReadyTalk: Conference code: 886-7778 Phone number: +1 (866) 740-1260 (U.S.A.)

Present: Herbert Greenlee, Jim Kowalkowski, Gianluca Petrillo, Brian Rebel, Saba Sehrish, Tingjun Yang

Access to database for MC events

The problem:

• database access is based on real time stamp from the event
• MC events don't necessarily have a valid and real time stamp

Should LArSoft provide specific facilities to support the MC access models the experiments plan to follow?

• Herbert Greenlee: MiroBooNE needs two modes
  1. “production” model: processing of MC events does not access any database; the information is statically stored in files and spans the whole experiment data taking
  2. “testing” model: MC events are given a valid time stamp and they access the DB, for example for testing the database itself, or specific scenarios
• Tingjun Yang: ArgoNeuT did use DB access for MC (once):
  • database access based on run number [equivalent, but not identical, to event time stamp (Ed.)]
  • run number distribution generated according to real data distribution
• Brian Rebel offered to check with Jonathan Paley for the model implemented by NOνA
• Herbert Greenlee reminded that realistic distributions will be obtained by overlaying data events with cosmic rays to simulated signal events; in this case, the event inherits the timestamp of the overlaid data event

No shortcoming has been identified in LArSoft's support for any of the discussed models.
The only identified requirement is the support for reading database-like information from a static data source. LArSoft database access design already supports this requirement.

Request of comments for changes of priority

Gianluca enumerated the coming aspects the review will touch:

1. implementation of the additional readout mapping in Geometry service
2. implementation of channel quality resource
3. factorization of DetectorProperties and LArProperties services

The conveened people do not have any request to modify this list.

Other business

It is expected that the next Architecture Review Meeting will happen in two weeks, on July 15th, 2015.

15:00 → 15:45 Discussion and new requests

=> how to check for gaps due to bad channels => timestamp-based database access for MC events => rising needs related to factorization? => current stoppers for interoperability? => any other business