Wrote mini scripts (complex commands in console, tomorrow I need to put them into the script; commands are in the file /mnt/t3nfs01/data01/shome/dbrzhech/history_bash_20180620.log) to analyse the batch jobs completeness: does a .root file was copied to the local area from a working node, does a .root file exists at my local path, etc. It is need because of huge number of running jobs, i.e the log files to analyse. So, with the help of these “mini scripts”, I've submitted the jobs that ran out of time (all.q queue for them was not enough, max time for all.q = 10h; the fraction of such jobs is not so big 38/~1100) and was not finished. I chose all.q yesterday as the fastest way to obtain results, because # of jobs that were running simultaneously in the long.q was 5-6 times less than in the all.q, but the number of files per job differs only by the factor of 3 (for long.q 15 files/job, for all.q - 5 files/job).

I found a mistake in my signal samples. I produced all the new signal samples: nominal and JEC(JES)/UP(DOWN) qq signal for the proper pt cut value (>40 GeV). However, I was confused with the naming of the nominal signal shapes and generated the intermediate mass points shapes with the old nominal signal shapes (with pt > 55 GeV). Fixed it! Copied the files with the shapes to the input folder of the combine tool. So, now I am sure that I am using the right signal shapes!

Tomorrow: check the jobs. If not all of them will be completed, run unfinished jobs again.

Fixed the mistake in the triplet code for the layer 2 residual distribution plots.

Tomorrow: I will check the fits and HV dependence. Check the z-dependence (residuals vs. z).

Reading about bias study to check the quality of fit.

Tomorrow: try to find this study in the dijet group tool and implement it in our analysis.