--
AmnonHarel - 19 Mar 2009
Introduction
See index in
AmnonHarelDataMixing
Testing
Mike's DataMixer, version V00-01-20. Goal was to see overlay work with active overlapping digis in data and MC (event overlays without overlapping digis look fine, see
AmnonHarelDataMixingRound4a).
Results
Identified a pair of event with significant overlap, and overlayed them. Worked the 3rd time (hence "special overlay #3").
Overlayed
MC event with
CRAFT event yielding
mixed event. The relevant digis (@ HB-, iPhi=40) are
MC digis with
CRAFT digis yielding
mixed digis.
Conclusions
- still not there. Further testing with Rick Wilkinson revealed that in his test areas the mixing works thanks to an old version of Mixing/Base. Rick then (April 6th) identified a change in the signature of BMixingModule's addPileups (in BMixingModule.h). There have been some discussion on which fix(es) are best.
- need debug prints of which event was overlayed on which event. Didn't Mike leave a flag to make a text file with this info?
- very annoying to not have a maxEvents in PoolSource
- can the user even control maxEvents for an SecSource?
- just generally anti-intuitive to have this parameter outside the PoolSource inputs.
How I got there
- work area:
/uscms/home/aharel/CMSSW_3_1_0_pre1/src/RecoJets/JetAnalyzers
- data area: contains both .log and .root files
/uscms_data/d1/aharel
- I started by trying to reuse the files from AmnonHarelDataMixing, but that crashed (the amazingly informative
px
0= assertion from boost shared_ptr) . Mike noted that completely consistent CMSSW versions are needed. So...
- I tried preparing the RAW MC with current (3_1_0_pre1) workarea with
prepMC03.py
. This gave a similar crash a bit later.
- Trying Mike's files... works
- Trying to overlay specific events:
- MC event: event #282 from mike's files,
mts3.py
- Non-collision data event: event #77 (i.e. run 68021, event #2669159) from mike's files
- picked out with
copyS3.py
--> S3event.root
- verified content with
dumpS3.py
--> dumpS3.root
--> the plots above.
- included in mix via
mts3_cfi.py
- Mixed event: recoed and dumped using
rds3.py
--> rds3.log
& rds3.root
--> plots above.
- Earlier attempts:
- MC event: event #282 from mike's files,
mts1.py
- Non-collision data event: event #77 (i.e. run 68021, event #2669159) from mike's files,
mts1_cfi.py
- since SecSources can't accept max events (maybe the global one gets there, hard to tell!), copied only that event with
copyS2.py
to S2event.root
- Mixed events:
mts1.root
- reco-ed with
recoMTS1.py
--> recoMTS1.log
& recoMTS1.root
- had to add my
recoMixed_310pre1_cff.py
and RecoLocalTrackerNoZeroSuppression_cff.py
, the latter needed to be updated with the ZDC inputs
- dumped with
dumpMixS1.py
--> dumpMixS1.root
. MC event #282 is in first event, can't tell which OD event is there...
- Mixed 1 event:
copyS2.py
(see above), mts2.root
, and mts2_cfi.root
- dumped with
rds2.py
---> wrong data event? ---> dumpS2.py
--> dumpS2.log
--> Yep. This is event #1 !? Oh, made a mistake. I sure hate not having maxEvents in the PoolSource where it so clearly belongs from the user's point of view.