Present: Vito Di Benedetto, Giuseppe Cerati, Lynn Garren, Hebert Greenlee, Alex Himmel, Thomas Junk, Katherine Lato, Gianluca Petrillo, Erica Snider, Hans Wenzel
Remote: Gleb Sinev, Jason Stock, Robert Sulej, others
Project Status Report [Erica Snider]
-
no feedback for LArSoft v6.10 with Geant4 10.2 and and GENIE 2.11.2
-
IFDHC 2.0 supports file search (see Vito Di Benedetto's talk), with a new syntax
-
proposing to deprecate BezierTrack
Conclusions:
-
we do need feedback on new Geant4 and GENIE
-
we also need feeback on the interest for Docker images
CI integration report [Vito Di Benedetto]
-
new DUNE test workflow is being tested
-
IFDHC 2.0 supports file search with
ifdh findMatchingFiles
instead of ifdh ls
Documentation report [Katherine Lato]
-
still needing more authors to document the algorithms and services they maintain
-
[Herbert Greenlee] automatic redmine e-mails get easily overlooked [A] we write direct e-mails when an issue is put in "feedback" state
-
following the model of TPC (SimChannel)
-
LArG4
will be extended
-
a new
BackTracker
like services
-
[Gianluca Petrillo] this code does not support truth extraction for full simulation [A] That's by design and acceptable; the design does not prevent such a feature from being added at a later time
Kalman fitter project [Giuseppe Cerati]
-
goal: write an optimal Kalman filter to fit a track
-
using MicroBooNE-like self-generated samples, would like to use official ones
-
establishing a reference performance record with three existing tracking algorithms
-
pull distribution for track start direction is a bit too narrow
-
pull distribution for track start position needs to be understood
-
the first version of new fitting algorithm (no pattern recognition) is encouraging
Data product review: phase II [Gianluca Petrillo]
-
[Erica Snider] having no indexing operator will be a potential issue
-
[Thomas Junk] referencing the coordinate names explicitly might make harder to have protoDUNE support with a different drift direction
-
[Thomas Junk] there are instances where time and position are stored in a single
TLorentzVector
in non-homogeneous coordinates, making silently invalid some of the operations on it; it would be better to have them stored separately
Conclusions:
-
would like to hear comments and concerns about this new library
There are minutes attached to this event.
Show them.