cmssnu / skflatanalyzer Goto Github PK

View Code? Open in Web Editor NEW

0.0 12.0 26.0 19.31 MB

Analyzer using SKFlatNtuple

C++ 14.50% Makefile 0.36% C 49.14% Shell 2.25% Python 33.75%

skflatanalyzer's Introduction

SKFlatAnalyzer

Manual

https://jskim.web.cern.ch/jskim/SKFlat/Manual/Manual_SKFlat.pdf

Where to put the analyzer

TAMSA 1/2 : /data6/Users/$USER/

KISTI : ~/ # home directory

KNU : ~/scartch/

First time setup

#### When first time gie clone, use the option "--recursive" to initiate the submodules
git clone --recursive [email protected]:CMSSNU/SKFlatAnalyzer.git
cd SKFlatAnalyzer
#### add your remote repo
git remote add <nickname> [email protected]:<gitaccount>/SKFlatAnalyzer.git
git checkout <your working branch>

#### First time setup script
source bin/FirstTimeSetup.sh 
source setup.sh

#### You have to edit user info
#### First, copy the temply using the command below
cp $SKFlat_WD/python/UserInfo_template.py $SKFlat_WD/python/UserInfo_${USER}.py 
#### Then, edit $SKFlat_WD/python/UserInfo_${USER}.py

#### Now, run setup script.
#### This should be done for every new shell
source setup.sh

Compile

make clean
make

Test job

SKFlat.py -a ExampleRun -i DYJets -n 50 -y 2016 &

Making a new Ananlyzer

cd script/MakeCycleSkeleton/

Then, run

python MakeCycleSkeleton.py NewAnalyzer # <- put new analyzer name

It will print below lines (execute the lines) :

mv NewAnalyzer.h $SKFlat_WD/Analyzers/include/
mv NewAnalyzer.C $SKFlat_WD/Analyzers/src/

Then, add

#pragma link C++ class NewAnalyzer+;

in Analyzers/include/Analyzers_LinkDef.h

Detailed descriptions

Look Analyzers/src/ExampleRun.C

Adding samples

To add a sample, you should add two files.

data/$SKFlatV/$ERA/Sample/ForSNU/$ALIAS.txt: list of file paths.
data/$SKFlatV/$ERA/Sample/CommonSampleInfo/$ALIAS.txt: alias, DAS name, cross section, nevent, sum(sign) and sum(weight).

And one file should be edited

data/$SKFlatV/$ERA/Sample/SampleSummary_*.txt: CommonSampleInfo files in one file. This file is not actually used by SKFlatAnalyzer. It is just a summary for users.

You can do it manually, or use scripts as below. The scripts use SampleSummary files for all SKFlat versions to find alias and cross section. If the sample is never used before the scripts will ask you alias and cross section. If it is annoying you can make temporal SampleSummary file like data/$SKFlatV/$ERA/Sample/SampleSummary_temp.txt which containing alias, DAS name and cross section (other information is not needed).

Make the file list file using bin/UpdateSampleForSNU.sh script.

./bin/UpdateSampleForSNU.sh $SAMPLEDIRECTORY

Run GetEffLumi analyzer to get nevent, sum(sign) and sum(weight)

SKFlat.py -a GetEffLumi -e $ERA -n 20 -i $SAMPLEALIAS

Make CommonSampleInfo file using bin/UpdateCommonSampleInfo.sh script.

./bin/UpdateCommonSampleInfo.sh

Update SampleSummary file using Summarize.py

cd data/$SKFlatV/$ERA/Sample
python Summarize.py

Tips

Making PR

Start from the CMSSNU's master branch of CMSSNU when making pull request to prevent anoying conflicts.

skflatanalyzer's People

Contributors

Watchers

skflatanalyzer's Issues

[BTag] FastSim to FullSim corrections updated for all years

https://hypernews.cern.ch/HyperNews/CMS/get/btag/1662.html

BTagSF for 2018 and 2016 available

https://hypernews.cern.ch/HyperNews/CMS/get/exotica/2688.html

This should be added.

Adding 2016 PU reweight

Adding PUreweight for 2016

Adding important analyzers into the master branch

I want to track analyzers which are used to obtain histograms for AnalyzerTools (e.g., Nvtx profile, MC b-tagging efficiencies, ... etc).
I don't see any harm to have them in the master branch, because they should be always able to be reproduced with any new sample production.
And they must be maintained followed by the updates of the master branch.

To @sungbinoh,
Can you add Analyzers/src/NvtxProfile.C in your private branch to the master branch? (this one)
To @jalmond
Can you add the analyzer you used to obtain MC b-tagging efficiencies?

[xsec]

Placeholder for #74

Adding BTagSF

Twiki : https://twiki.cern.ch/twiki/bin/view/CMS/BtagRecommendation#Recommendation_for_13_TeV_Data

Job directory size is large

Now, when submitting batch jobs, we copy libraries (and source codes) to each samples.
E.g.,

However, they all use the same files if they were submitted with one SKFlat.py command (with -l samples.txt).

So, we can try

|-- ExampleRun__2019_02_12_073246__Year2017
| |-- <libraries and source codes>
| |-- WW
| |-- WZ
| |-- ZZ

But I think it's better to start this after we change our schedulers of tamsa to condor.
Until then, users must always delete them regularly.

bool Jet::IsTagged(Tagger tg, double wp){
  double tr = GetTaggerResult(tg);
  return (tr>wp);
}

or should we think more clever way?

(https://github.com/CMSSNU/SKFlatAnalyzer/blob/master/DataFormats/src/Jet.C#L126-L180)
(https://github.com/CMSSNU/SKFlatAnalyzer/blob/master/DataFormats/src/FatJet.C#L140-L194)

Moving Run2UltraLegacy to master

As many of use using UL branches, maybe it is time to move Run2UltraLegacy branch to master branch and old master (Run2Legacy) to another separated branch. Any opinion?