Present: Jim Amundson, Brandon Eberly, Laura Fields, Lynn Garren, Herbert Greenlee, Jim Kowalkowski, Thomas Junk, Jonathan Paley, Marc Paterno, Gianluca Petrillo, Saba Sehrish, Mitchell Sodeberg, Michelle Stancari, Karl Warburton, Hans Wenzel, Tingjun Yang, Joseph Zennamo
Connected: David Adams, Vito Di Benedetto, Gleb Sinev
Project Status Report [Gianluca Petrillo]
-
LArSoft v04_23_00 was released last week (including fixes on geometry,
MCShower -- but buggy needs to be fixed)
-
[Herbert Greenlee] what was wrong? bug arose with the addition of
AuxDetGeometryService: after then, geometry specified in input files would not be loaded automatically any more
-
this week release: bug fix to
MCShower, Tingjun/Xin/Chao’s WireCell
-
[Lynn Garren] an e-mail is circulating in LArSoft mailing list detailing the changes; unless replies to that e-mail advise against it, we'll berge the feature branch
-
architecture review update:
-
regular Wednesday meetings
-
ongoing work: channel filtering services and LArProperties
-
plan is to factorize the core services, 5-6 services
-
[Jonathan Paley] pointing out that he is heavily modifying TimeService
-
[Jonathan Paley] LArFFT? it is in the queue to be factorized (by Gianluca)
-
work on modules after wrapping up the services
Remarks: Tingjun/Xin/Chao’s WireCell feature branch will be merged into the coming release unless replies by e-mail advise against it.
Channel Status service [Brandon Eberly]
-
use of "provider and service" model established for pedestal DB readout
-
created 2 implementation: source from FHiCL configuration or from database
-
ChannelFilter (legacy code, deprecated) to be a wrapper around the new service
-
channel filter service interface - channel statuses must always be implemented
-
[Gianluca Petrillo] legacy behaviour: channels are tagged good if they are not bad nor noisy and are present
-
[David Adams] is wire-channel mapping provided here or elsewhere? wire-channel is provided by Geometry service [aside: (David Adams) aspecific service might be created for it]
-
[Jonathan Paley] do channels added on a event by event basis with AddNoisyChannel() disappear on the next event? when do they get cleared? not sure, probably they get clear on new event
-
other changes:
-
ChannelID changed from uint64_t to uint32_t
-
interface uses types from pedestal interface
-
feature branches are ready to be merged
-
for the future, data product might be needed to store noise channel status per event
-
no time to discuss the Organization/Naming slide... next time
Changes needed before merge:
-
renaming of the service to a final name (possibly
IChannelStatusXxx)
-
update of FHiCL configuration of all experiments
Why you should upgrade to Geant 4.10 [Hans Wenzel]
-
new Geant4 release, 4.10.01_p02, released on June 19th
-
support for multi-threaded Geant4 (event level)
-
physics example from what LArIAT cares for:
-
electron vs. photon shower discrimination, muon sign determination
-
changes affecting physics:
-
all 6 electromagnetic variants are available (see slides)
-
if interested in kaons, the old version of Geant4 should not be used
-
expect 5% speedup
Experiments are invited to internally discuss about their plan with Geant4.
MCShower [Joseph Zennamo]
-
improved momentum computation: direction of shower computed within 2.4 cm diameter, then calculate momentum
-
dE/dx and dQ/dx calculations added/fixed
-
"2.4 cm" is hard coded: it should be turned into a parameter in the future
No action is required, since the code is already in develop branch.
A deep muon generator [Karl Warburton]
-
propagates muons deep underground to calculate backgrounds
-
code is currently residing in a branch feature/php13tkw_GaisserParam in larsim (up to date with develop)
-
code is ready to be merged (minor tweaks needed?)
No objections have been raised. The code will be merged in the next release.
CI Test in LArSoft [Vito Di Benedetto]
-
LArSoft regression test suite implemented
-
published in branches feature/vdb_ci_regression_test_suite for all experiment repositories (argoneutcode, dunetpc, lar1ndcode, lariatsoft, uboonecode)
-
all tests run well within the target time upper limit (10 minutes)
-
[Thomas Junk] the memory used by MicroBooNE's test in G4 phase is dangerously close to 2 GB
-
[Gianluca Petrillo] how are the data files compared? currently, only the appearance or disappearance of a data product from the file is considered an error
-
[Gianluca Petrillo] it is in the plan to run this in Jenkins, to allow automatic testing
There are minutes attached to this event.
Show them.