Running Analysis on non-generated Samples/ No procDict

FCCAnalysis
1

Hello,

At the moment I am trying to generate and analyze a BSM process that is specified in my pythia card but not in a process in the procDict. The old FCC starterkit tutorial was generalized, but now using the ‘fccanalysis’ command the process requires that the analysis file be run over a process specified in the process dictionary. Is there a way to apply the analysis files to non-procDict defined events?

Thanks,
Hayden

2

Hello @hshaddix,

I believe you will have to create additional entry to the procDict using procDictAdd variable like so:

procDictAdd={"MySample_p8_ee_ZH_ecm240":{"numberOfEvents": 10000000, "sumOfWeights": 10000000, "crossSection": 0.201868, "kfactor": 1.0, "matchingEfficiency": 1.0}}

More details is in the example analysis here.

best,
Juraj

3

Hey @jsmiesko ,

When I do that and try to run, I then should not use prodTag? I am unsure of how to set an inputDir or prodTag which are required if I set my own process.

Thanks,
Hayden

4

Hello @hshaddix ,

if you want to use also existing samples, yes you need something like:

Cheers,
Clement

5

@clement.helsens ,

When I add a new process to the first stage of analysis using the few lines:

procDictAdd= {‘p8_ee_HA3_ecm240_edm4hep’:{‘numberOfEvents’:100000,‘sumOfWeights’:100000,‘crossSection’:0.001,‘kfactor’:1.0,‘matchingEfficiency’:1.0}}
processList = {
‘p8_ee_HA3_ecm240_edm4hep’:{}
}

I am adding my specific event from the generation step using DelphesPythia8_EDM4HEP and then trying to add the process to the procDict so I can run the analysis over it, but when I do it tells me ‘attribute has no namespace ‘final’’ implying that the lines I added should be run in the analysis_final.py file instead, I must be misunderstanding what it is asking.

Thanks,
Hayden Shaddix

6

hello @hshaddix ,

indeed, the procDictAdd is only possible in the final stage of the analysis like fccanalysis final examples/FCCee/higgs/mH-recoil/mumu/analysis_final.py

this final stage is producing normalised histograms, thus it needs to know the sample cross section.
Hope this helps,
Clement