Fixing typos and integration of mkPlot when working with envelope nuisances #18
Conversation
…ces and works fine
…in name to the subsample name...This is really hardcoded. If One lift this then the user is specified to user whatever name in the subsample in samples.py and that I feel is more user firendly
A more general install script dependent on the operative system
| for sub in sample["subsamples"]: | ||
| samples["%s_%s" % (sname, sub)] = sample | ||
| subsamplesmap[-1][1].append("%s_%s" % (sname, sub)) | ||
| samples["%s" % (sub)] = sample |
There was a problem hiding this comment.
Not sure what the problem is, this is coming from original LatinoAnalisys code here
There was a problem hiding this comment.
The code will always prepend the "sample" name to the "subsample" name. Suppose you have a sample, and you split into subsamples with different event weight then the subsample name is not user defined and that could affect other part of the process, like tayloring the datacrds to combine models (e.g you have a sample "Zee" and you want it into the "sm" subsample, "quad_cW" subsample, .... you'll never be able to have the correct naming).
By lifting this the user has full flexibility on the subsamples names and can alsp prepend the sample name if he wants to
There was a problem hiding this comment.
Wouldn't the final names be ['Zee_sm', 'Zee_quad_cW'] ?
There was a problem hiding this comment.
With the old code yes. With the new one it would be "sm" and "quad_cW" There is no way to get rid of the initial Zee_ in the old setup (if not using something external to mkShapesRDF) which is clearly unnecessary.
There was a problem hiding this comment.
Cannot support 3 different LCG releases. What was done for the master branch of latinos/mkShapesRDF was to stick with 103 for all the OS (see here).
Also el8 should be a unusual OS for the machines we work with.
There was a problem hiding this comment.
Ok then this should be integrated also here at some point.
I understand el9 is the standard but if the code is olnly made to work on el9 machines one should clearly specify it in the instruction at the early stages
2 participants
This PR fixes some typos and better integrates mkPlot with ongoing developments in mkShapesRDF concerning the handling of envelope nuisances.
It is suggested to make the "flow" syntax of hist2array less error prone.
The proposed suggestions were tested in a simplified environment involving LHE nanoAOD supporting QCD and PDF variations: A simple observable both with fold=0 and fold=3. The shapes were both produced when running the code locally and via batch submission.
In order for the processing of envelope nuisances to work locally, some fixes had to be done.