PR to clean up the code #16
Conversation
| @@ -233,7 +231,7 @@ default: | |||
| regionregex: .* | |||
| apply_hf_cuts: True | |||
| hf_pt_thresh: 80 | |||
| region_without_dijet_cuts: False | |||
| region_without_dijet_cuts: True | |||
There was a problem hiding this comment.
Do we want to keep this on by default? I think it's fine, I was just thinking it might introduce additional runtime + memory constraints. Especially removing the dijet cuts can increase the number of events passing quite a bit - hence making ParticleNet input dict occupy larger amounts of memory.
There was a problem hiding this comment.
let's keep this on hold. I think that we should check the DNN distributions without these cuts.
bucoffea/vbfhinv/definitions.py
Outdated
| regions.update(tmp) | ||
| to_add = ['met_sr', 'mindphijm'] | ||
| to_remove = ['veto_muo', 'veto_tau', 'veto_ele', 'mindphijr', 'recoil'] | ||
| regions.update(dict([(f"{region}_no_veto_all", add_lists(clean_lists(regions[region], to_remove),to_add)) for region in regions.keys() if region.startswith("sr_")])) |
There was a problem hiding this comment.
I love the one-liner, but it might be confusing for new people when looking at the code to understand what this does I think :) What do you think? Can you maybe at least make the for loop explicit like the previous version? Or if you want to keep it this way, maybe a comment on top explaining what this does might be helpful.
There was a problem hiding this comment.
Done. Let me know if you want some more clarity more
| if not os.path.exists(outdir): | ||
| os.makedirs(outdir) | ||
| outdir = pjoin('./output/',args.inpath.replace('..','').replace('/',''),'hf_estimate') | ||
| dump_info(args, outdir) | ||
|
|
||
| outrootpath = pjoin(outdir, 'vbfhinv_hf_estimate.root') |
There was a problem hiding this comment.
Quick question: If I understand this correctly, you're saving the file under output/{outtag}/hf_estimate, and in the above script (where plot_data_mc function is defined) I think you're reading the file from {outtag}/hf_estimate. Did I get that right? Just wanted to check if it's working :)
There was a problem hiding this comment.
the folder should be the same in the end. maybe the naming convention could do a bit of polishing
…into particlenet Conflicts: bucoffea/vbfhinv/vbfhinvProcessor.py
No description provided.