Upload
others
View
26
Download
0
Embed Size (px)
Citation preview
News and IssuesTom JunkDUNE Software Management MeetingDecember 20, 2017
Upcoming Dates• Holidays 2017- Would like to use the time to run MCC10 – 6M FD neutrino events
requested by deep learning teams
• January 29 to Feb. 1, 2018: Collaboration meeting at CERN- Expect two S&C parallel sessions, possibly shared with related
groups
- Lunchtime tutorials
• Feb. 2,3 ProtoDUNE Data Readiness Review at CERN- Addresses computing resources, reco algorithms, OM, DQM,
calibration, offline operations
12/20/17 Tom Junk | News and Issues2
Pandora and 35-ton• Redmine issue:
https://cdcvs.fnal.gov/redmine/issues/18584• Easy problem – fcl files needed updating- particlestitcher functionality now in standard Pandora module, so no
need to include it as a separate step.
- Small amount of rewiring module labels so data still flows.- John Marshall provided an example and it was easy to follow (once
we knew what to do).
• More difficult problem: larpandoracontent/LArPlugins/LArRotationalTransformationPlugin.ccemits this message when running with 35t and its caller throws an unknown exception:LArRotationalTransformationPlugin::Initialize - Plugin does not support provided LArTPC configurations
12/20/17 Tom Junk | News and Issues3
Pandora and 35-ton cont'd• Pandora team discussed this. The current plugins assume the
same wire angles in different drift volumes, and 35t breaks this assumption. Two options presented in the ticket:
1. We have tested and pushed a branch "feature/LArRotationalTransformationPlugin" to Redmine that will allow the DUNE35t reconstruction to proceed, with the understanding that the transformations will only be approximations in some drift volumes. The diff should read cleanly, and will show that the maximum allowed difference between like wire (u and v) angles between volumes is now configurable and carries a larger default value.
2. We decide that Pandora should no longer be included in the DUNE35t reconstruction, as the Pandora team does not have sufficient person power to maintain it. It is not an experiment that Pandora has targeted directly, but has been able to use developments put together for e.g. MicroBooNE. Our routine testing considers tens of thousands of MicroBooNE events nightly, about ten thousand ProtoDUNE events upon demand and a few hundred DUNEFD events upon demand. It does not include DUNE35t (more detail about our tests, with Travis CI, Coverity, codecov and Valgrind are available upon request).
Please let us know which you prefer.
12/20/17 Tom Junk | News and Issues4
Discussion of Options• If we were still considering 45±εo for the FD wire angles, we
would dig our heels in on this and insist that a plugin be created.• But as it is, 35-ton data analysis is almost complete, and• The publications aren't using Pandora reco• We could keep Pandora for the long drift volume, and live with
approximate geometry for the short drift volume.• It would make analyses confusing if the short and long drift
volumes were treated differently in this way however.
12/20/17 Tom Junk | News and Issues5
Muon tracking efficiency in Pandora• Tingjun noticed in the CI muon tracking efficiency test (1x2x6
FD workspace, GENIE events), that the muon tracking efficiency for Pandora drops from 96% to 72% with the Pandora release that comes with LArSoft v06_60_00
• Ongoing investigation – trying to reproduce in different environments or track down the issue.
• Update from yesterday evening – John Marshall was able to reproduce the issue and has committed a fix to a feature branch of larpandoracontent, asking Andy Blake to check it.
12/20/17 Tom Junk | News and Issues6
12/20/17 Tom Junk | News and Issues7
TRACEv3_08_01
artv2_08_04
canvasv3_00_03
clhepv2_3_4_4
cppunitv1_13_2
messagefacilityv2_01_01
pythonv2_7_13d
rootv6_10_04d
artdaq_corev1_07_12
awscliv1_7_15
boostv1_64_0
caffev1_0e
glogv0_3_5
gslv2_3
hdf5v1_10_1b
lapackv3_7_1
leveldbv1_20
lmdbv0_9_21
opencvv3_3_0a
protobufv3_3_1
gflagsv2_2_1
snappyv1_1_7
fhiclcppv4_06_01
tbbv2017_7
sqlitev3_16_02_00a
fftwv3_3_6_pl2
libxml2v2_9_4b
pythiav6_4_28h
mysql_clientv5_5_54
postgresqlv9_5_6
xrootdv4_5_0c
cetlibv3_01_01
cetlib_exceptv1_01_03
cigetcertv1_16_1
cigetcertlibsv1_1
setpathv1_11
cpnv1.7
cryv1_7h
dune_pardatav01_26_00
dunetpcv06_60_00_02
duneutilv06_60_00
genie_phyoptv2_12_8
genie_xsecv2_12_8
larsoftv06_60_00
lbne_raw_datav1_04_17
larbatchv01_34_03
ifdh_artv2_04_02
laranav06_13_02
larrecov06_49_00
lareventdisplayv06_15_05
larexamplesv06_07_13
larpandorav06_18_00
larsoft_datav1_02_01
larwirecellv06_08_07
jobsub_clientv1_2_5
sam_web_clientv2_0
eigenv3_3_3
geant4v4_10_3_p01a
geniev2_12_8a
lhapdfv5_9_1h
log4cppv1_1_2
pdfsetsv5_9_1b
ifbeamv2_1_3
ifdhcv2_2_3
libwdav2_24_0
ifdhc_configv2_2_3
nucondbv2_1_3
jsoncppv1_7_7
jsonnetv0_9_3
nutoolsv2_16_09
larsimv06_38_00
tensorflowv1_3_0a
larcorev06_15_08
larcorealgv1_12_00
larcoreobjv1_18_00
lardatav06_33_02
lardataobjv1_24_00
rangev3_0_2_4
nusimdatav1_08_04
larevtv06_16_12
marleyv1_0_0a
larpandoracontentv03_09_00
pandorav03_07_00
wirecellv0_6_2
dunetpc dep. tree with version numbers
gcc,cetbuildtools,andGEANT4'sdependenciessuppressed.
madefromoutputofupsdepend
Adding Unpacking modules
12/20/17 Tom Junk | News and Issues8
• Jingbo has added a feature branch to dunetpcfeature/online_monitoring_artdaq
• Has raw decoder modules for RCE, FELIX, SSP, and Timing data.
• We would like to include these in dunetpc's released version- introduces a dependence on dune-raw-data and dunepdsprce
- dune-raw-data is not yet included in prebuilt releases (built from source for the online).
• Feature branch was based on an older dunetpc with dependence on an older version of artdaq_core. This must be kepto stable so that the online TransferInput source works between artdaq and the OM processes.
Adding Unpacking modules• Development head of dunetpc can be setup along with dune-raw-data using
the "offline" qualifier: nu:e14:s56 (John and Tom verified this).
• dune-raw-data has two other qualifiers to maintain stable dependencies- online:e14:s50
- online_monitoring:e14:s50
• John Freeman (or Kurt and others) would have to maintain dune-raw-data versions moving forwards (long after ProtoDUNE-SP runs). Similar to lbne_raw_data, which is still on the maintenance list.
• John had supplied tarballs of lbne_raw_data to be unwound in /grid/fermiappand CVMFS. lbne_raw_data is also built on Jenkins as dunetpc is build on Jenkins and needs it.
• But then Jingbo's modules would have to be merged in to dunetpc on the develop branch. They may evolve independently on the feature branch.
12/20/17 Tom Junk | News and Issues9