WIP Add SV Fit #182

wants to merge 25 commits into

2 participants


This is a work-in-progress for adding SVFit to FinalStateAnalysis.

Shalhout Sha... and others added some commits Nov 21, 2012
Shalhout Shalhout First commit of FSA code with SVFit 1c86c11
Shalhout Shalhout Cached SVfit output to avoid calling multiple times for identical arg…

ZH candidate events with more than 1 SVfit call are found to have more than 4 leptons.
Each SVfit call is based on a unique choice of which leptons are 1-4.
Duplicate SVfit output is obtained when leptons 2 and 3 are fixed in a choice of leptons 1-4.
Caching the results avoids calls in this case.
@ekfriis ekfriis SVFit from Shalhout 89c112e
@ekfriis ekfriis Indentation fbfe123
@ekfriis ekfriis Remove extraneous class wrapper 0c81c56
@ekfriis ekfriis Refactor SVfit code
Remove dependency on TauAnalysis from DataFormats.

Cache the collections using the Hash.h features.
@ekfriis ekfriis Merge branch 'master' into svfit
@ekfriis ekfriis Don't depend on the variable when nuking 7d11a8c
@ekfriis ekfriis Use official forward decl. to prevent conflics 8dfd14a
@ekfriis ekfriis Fix major compile errors in SVfit algo 31739ff
@ekfriis ekfriis Fix type errors b16f68b
@ekfriis ekfriis Style cleanup f98e0d4
@ekfriis ekfriis The one true brace style 91a6045
@ekfriis ekfriis Don't emit error if user doesn't type 'n' 7c79e44
@ekfriis ekfriis Add logic to ntuple builder to only SVfit legs where it makes sense 35dd18d
@ekfriis ekfriis We need to always checkout the full PAT recipe to prevent segfaults.
This is because the selectors we use from PatAlgos are still linked
agains a very old dataformat in the full release.
@ekfriis ekfriis fix bug in CLI 61d1c8c
@ekfriis ekfriis Add MVAMET as a recipe option 55fd3e5
@ekfriis ekfriis Use MVA MET for SVFit 0e3f09f
@ekfriis ekfriis Crash gracefully if MVAMET is missing b7ca31e
@ekfriis ekfriis Use correct name for MVA met 93c4866
@ekfriis ekfriis Make running SVfit optional fa12e91

OK, @scooperstein I think this will work now. If would try to check this development branch out in a separate area and try to build the ntuples.

# Follow the install instructions here 

except at the clone & recipe steps:

# Get the SVfit development branch (i.e. this pull request)
git clone -b svfit https://github.com/uwcms/FinalStateAnalysis.git

# install with necessary dependencies
cd FinalStateAnalysis/recipe
PATPROD=1 MVAMET=1 ./recipe.sh
# compile etc

then try to run the limits

cd FinalStateAnalysis/NtupleTools/test

submit_job.py 2013-04-07-8TeV-v1-ZH make_ntuples_cfg.py channels=zh rerunFSA=1 rerunMVAMET=1 svFit=1 --tuple-dirs=$fsa/MetaData/tuples/PATTuples-8TeV.json --apply-cmsRun-lumimask --input-files-per-job=1 --shared-fs  --samples "*VH*" "*WZ*" "*Zjets*" > do_higgs.txt

bash < do_higgs.txt

It should produce new ntuples in your HDFS area in the folders 2013-04-07-8TeV-v1-ZH/SAMPLE_NAME with the SVfit, for the VH, WZ, and Zjets samples. Can you try and run this and see if it works?

Stephane Coo... and others added some commits Apr 18, 2013
Stephane Cooperstein (CMS svfit 8ed4aac
Stephane Cooperstein (CMS fixed small error in tau template 17965e7
@ekfriis ekfriis Merge pull request #190 from scooperstein/svfit
Fixed a small error in tau template, update some data locations.
@ekfriis ekfriis referenced this pull request Jun 20, 2013

Add SVFit functionality #226


Superseded by #226

@ekfriis ekfriis closed this Jun 20, 2013
@ekfriis ekfriis deleted the svfit branch Jun 20, 2013
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment