ra4_draw

Repository for plotting and table-making utilities combining the flexibility of analysis_code's simple and flexible string-based plotting with the speed of ra4_macros' "looping" scripts.

Commmands for RA4 analysis plots

To get just the pure MC kappa plots:

./compile.py && ./run/ra4/kappa_plots_tables.exe --mm off -m signal --mc_kappas -y 0 -n

Check kappa values in 5-6 jet CR, 2l+veto CR and SR+lowMET using MC in place of data:

./compile.py && ./run/ra4/kappa_plots_tables.exe --mm data -m cr56j --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm data -m cr2lveto --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm data -m lowmet --data_kappas -y 0

Unblinding SRs:

./compile.py && ./run/ra4/kappa_plots_tables.exe --mm data -m signal --data_kappas -y 0 -u

Additional options:

* Use flag `--data_kappas` to make the kappa plots comparing mc to data or mc to mc w/ data stat
* Use flag `--mc_kappas` to make the kappa plots showing just nominal MC kappas
* Add the flag `--debug` (or `-d`) to see the cuts for each region. 
* Use `--mm data` to compare actual data to MC in the CRs. Use `--mm off` to use MC as pseudodata.
* Add the flag `--unblind` (or `-u`) to plot data in SRs as well.

For partial kappa plots in the systematic section:

./compile.py && ./run/ra4/plot_partial_kappa.exe -t lowmj -y 0 # plot all bkg low-MJ
./compile.py && ./run/ra4/plot_partial_kappa.exe -t highmj -y 0 # plot all bkg high-MJ
./compile.py && ./run/ra4/plot_partial_kappa.exe -t highmj -y 0 -n 1 -d 1 # plot wjets high-MJ
./compile.py && ./run/ra4/plot_partial_kappa.exe -t highmj -y 0 -n 2 -d 2 # plot ttjets high-MJ

Plots showing that CRs mimic behavior of SRs:

./compile.py && ./run/ra4/kappa_plots_tables.exe --mm mismeas_kappa -m cr56j_metbins --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm mismeas_kappa -m cr2lveto_metbins --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm mismeas_kappa -m signal_metbins --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm w_isr -m cr2lveto_njbins --data_kappas -y 0
./compile.py && ./run/ra4/kappa_plots_tables.exe --mm w_isr -m signal_njbins --data_kappas -y 0

Generate a specific datacard, or wihtout signal uncertainties for the results plot:

./compile.py && ./run/ra4/write_datacards.exe -p 2100_100,1900_1250 -y 0 -u

Getting the detailed signal systematics tables:

./python/sig_sys_table.py systable.tex sys_SMS-T1tttt_mGluino-2100_mLSP-100_0_nom.txt sys_SMS-T1tttt_mGluino-1900_mLSP-1250_0_nom.txt

Get limits, this also generates the datacards for the full scan:

./python/send_limits.py

Limit plot using the concatenated text files from previous step:

./compile.py &&  ./run/ra4/limit_scan.exe -t nom -f txt/limits/limits_t5tttt_run2.txt -s 0 &&  ./run/ra4/limit_scan.exe -t nom -f txt/limits/limits_t1tttt_run2.txt -s 0 && ./run/ra4/plot_t5tttt_prl.exe 

To perform the fit without the signal region observations, start by creating a workspace in order to create the masking variables:

text2workspace.py datacard_SMS-T1tttt_mGluino-2100_mLSP-100_0_none.txt --channel-masks
text2workspace.py datacard_SMS-T1tttt_mGluino-1900_mLSP-1250_0_nom.txt

The resulting workspace can be examined using dump_workspace.cxx. Then, to do the fit with masked R4s:

./compile.py && ./run/ra4/write_datacards.exe -p 2100_100,1900_1250 -y 0 -u --no_syst

combine datacard_SMS-T1tttt_mGluino-2100_mLSP-100_0_none.root -M FitDiagnostics --forceRecreateNLL --saveWorkspace --saveWithUncertainties --saveOverall --setParameters=mask_r4_lmet_lnb_lnj_lmj=1,mask_r4_lmet_lnb_lnj_hmj=1,mask_r4_lmet_lnb_hnj_lmj=1,mask_r4_lmet_lnb_hnj_hmj=1,mask_r4_lmet_mnb_lnj_lmj=1,mask_r4_lmet_mnb_lnj_hmj=1,mask_r4_lmet_mnb_hnj_lmj=1,mask_r4_lmet_mnb_hnj_hmj=1,mask_r4_lmet_hnb_lnj_lmj=1,mask_r4_lmet_hnb_lnj_hmj=1,mask_r4_lmet_hnb_hnj_lmj=1,mask_r4_lmet_hnb_hnj_hmj=1,mask_r4_mmet_lnb_lnj_lmj=1,mask_r4_mmet_lnb_lnj_hmj=1,mask_r4_mmet_lnb_hnj_lmj=1,mask_r4_mmet_lnb_hnj_hmj=1,mask_r4_mmet_mnb_lnj_lmj=1,mask_r4_mmet_mnb_lnj_hmj=1,mask_r4_mmet_mnb_hnj_lmj=1,mask_r4_mmet_mnb_hnj_hmj=1,mask_r4_mmet_hnb_lnj_lmj=1,mask_r4_mmet_hnb_lnj_hmj=1,mask_r4_mmet_hnb_hnj_lmj=1,mask_r4_mmet_hnb_hnj_hmj=1,mask_r4_hmet_lnb_lnj_lmj=1,mask_r4_hmet_lnb_lnj_hmj=1,mask_r4_hmet_lnb_hnj_lmj=1,mask_r4_hmet_lnb_hnj_hmj=1,mask_r4_hmet_mnb_lnj_lmj=1,mask_r4_hmet_mnb_lnj_hmj=1,mask_r4_hmet_mnb_hnj_lmj=1,mask_r4_hmet_mnb_hnj_hmj=1,mask_r4_hmet_hnb_lnj_lmj=1,mask_r4_hmet_hnb_lnj_hmj=1,mask_r4_hmet_hnb_hnj_lmj=1,mask_r4_hmet_hnb_hnj_hmj=1 --numToysForShapes 10000 --name=_nor4_2100

combine datacard_SMS-T1tttt_mGluino-2100_mLSP-100_0_none.root -M FitDiagnostics --forceRecreateNLL --saveWorkspace --saveWithUncertainties --saveOverall --numToysForShapes 10000 --name=_r4_2100

combine datacard_SMS-T1tttt_mGluino-1900_mLSP-1250_0_nom.root -M FitDiagnostics --forceRecreateNLL --saveWorkspace --saveWithUncertainties --saveOverall --name=_r4_1900

Repeat w/o the --setParameters argument to get the global fit including R4's. The result plot can then be made using the output files from the above fit with the python/plot_results.py script.

Goodness of fit estimate:

combine -M GoodnessOfFit datacard_SMS-T1tttt_mGluino-2100_mLSP-100_0_nom.txt --algo=saturated --fixedSignalStrength=0
combine -M GoodnessOfFit datacard_SMS-T1tttt_mGluino-2100_mLSP-100_0_nom.txt --algo=saturated --fixedSignalStrength=0 --toysFreq -t 100

Then check how many of the toys are above the value gotten from the data in the first command in the "limit" branch.

Code documentation

Doxygen-based documentation is available at Adam's UCSB-HEP webpage.

Setup and compilation

Compilation requires c++11 and ROOT, but not CMSSW. To compile, simply run

./compile.py

Making histograms

An example script is available under src/test.cxx. To execute, compile and then run

./run/core/test.exe