Daily Log 2013
August
20
checking memory leaks
Found this tool (
valgrind) for checking memory leaks amongst other things. Brian expressed some concern over leaks in the ntupleProducer code. I'm not sure how to interpret the results, but this seems to work for getting whatever info valgrind provides,
valgrind --tool=memcheck `cmsvgsupp` --leak-check=yes --show-reachable=yes --num-callers=20 --track-fds=yes cmsRun your_cfg.py >& vglog.out &
Just running over 100 events produced a log file that is 202 MB!
12
b-tagging mc truth
This twiki has not been updated (okay, part of it was) forever, and doesn't even have a fully hashed out example of how to do this without
PAT,
https://twiki.cern.ch/twiki/bin/view/CMSPublic/SWGuideBTagMCTools. I could use what Radek was doing... His most recent work seems to be here, ~radek/HZZAna3/CMSSW_4_4_4/src/DPSAna/ntupleProducer.
July
3
Questions for Pedro
- What trigger efficiencies are being used and are analysis objects (leptons) matched to trigger objects?
- Tools for jet flavor location...
- Can we use b-flavor content probablility as in the Vtb measurement as a discriminator between FCNH and ttH final states?
June
27
Most of the codes have been ported to github!!! One useful thing is to exclude certain files. This is done by making a file .gitignore_config with a list of suffixes that should be ignored when running
git add
. For example, to ignore compiled code, include something like this,
# Compiled source #
###################
*.com
*.class
*.dll
*.exe
*.o
*.so
*.pyc
To make this part of your global configuration, run the following command,
git config --global core.excludesfile ~/.gitignore_global
21
FCNH
- 20130621_162756: 2012AB data; testing removal of lepton scale factors. Not a huge difference between here and here. I'm surprised by having what appears to have a small acceptance for triboson production in 3 lepton final states. This makes sense if the MC samples tend to be dominated by low pt leptons. Probably the case...
May
31
lumi/BRM
- Need to write up some instructions for the VdM simulation code. Just need an overview with some simple running instructions for Esha.
- Beamspot info is included. Need to format plots and get 2D fit output to be compatible with bias method
- Also need to include difference between widths measured from fits and width of fit function.
FCNH
Need to take stock of where I'm at...
April
12
FCNH
Made some changes to the cuts:
- Z veto is done for both OSSF dilepton pairs and trileptons, i.e., |Mlll/OSSF - MZ| < 7.5
- MET and HT cuts are applied simultaneously. Needs to be optimized still. For now I require [MET > 20 and HT > 50 and ((1.5*MET)2 + HT2) > 1002]
- Instead of requiring Nb-jet > 1, I'm looking at the 0 b-jet bin and cutting on the #Sigma discrJPB of the anti-tagged jets (discrJPB < 0.275).
March
28
FCNH
Fixed a bug with batch system: was trying to copy files from soft linked 'histos' directory. Was not working. Reverted back to old system.
FCNH sample with H->bb generated and published.
27
FCNH
Checking out some
2D distributions to check correlations (signal on top, bg on the bottom) between known discriminating variables. Producing some Dalitz plots (i.e., M
12 vs. M
23), but they're still running.
Comments from Mayda
Check out event display for events in z window for M_lll
- Check out clean-up filters (cosmics, muon halo, etc.). Do they kill events.
- Check effect of MC hlt selection (not currently applied)
- redo trilepton mass comparison with ZZ->4l samples from madgraph and powheg, apples-to-apples
- Consider events with M_OS > 85
- Investigate effect of MET + HT cut on deltaR VS M_OS
22
FCNH
From Mayda
- Investigate dilepton dr plot more. Add to note and focus on first couple bins with finer binning.
- 4l mass. Read 4l PAS (may be in a talk) on rescaling to match Z peak.
- Look at discriminating power of jet + HT requirement versus b-jet requirement or something along these lines
21
FCNH
Current selection (as in note). 5_trilepton_mass_window directory has plots where only events in with |M
lll - 90| < 10. The goal is to see if there is anything obviously amiss in other distributions for the ll+μ categories. There is a clear deficit for this mass window in said events and could be coming from Zbb->llbb events.
Running over Zbb, Wbb, and t-channel single top to check if these deal with excesses seen in ll+μ categories and deficits in same-sign selections.
Adding jet energy resolution scale factors as described
here. I'm using the first method with gen jet matching.
8
FCNH
Preparing sync with WH and presentation for Thursday. Sync plots for WH analysis are in fcnc_2012/20130308_122811. Need to be redone: the top veto was applied twice so it's not counted for the results tables.