ValidateDatacards.py Problem: Systematic type autoMCStats not supported about combineharvester HOT 4 CLOSED

Hi @okarache,
Can you share the file shapes_WPhiPS50mu_3L_LS0_Run2.root so we can try to reproduce your problem? I've never seen this before. In parallel, can you paste the commit hash printed at the top when you do git log in the CMSSW_10_2_13/src/CombineHarvester

from combineharvester.

Hi, @adewit ,

thank you for you response.
I uploaded the shapes_WPhiPS50mu_3L_LS0_Run2.root to cernbox here: https://cernbox.cern.ch/index.php/s/Q5nPehbbMEKPz6O

Here is git log printout:
/cms/multilepton-3/olena/CMSSW_10_2_13/src/CombineHarvester$ git log
commit 0675d6a (HEAD -> master, origin/master)
Merge: 3efe0f2 5f16a9c
Author: Andrew Gilbert [email protected]
Date: Wed Jul 21 17:06:29 2021 +0200

Merge pull request #260 from threiten/dev_skipInitialFitImpacts

Added option to skip running initial fit for impact plots.

Thank you and with best wishes,
Olena

from combineharvester.

Hi @okarache, thanks for reporting this - it's an interplay between the number of processes you have in the datacard, which means that the normal systematics lines are exactly as long (in terms of number of items) as the full autoMCStats line, leading to the wrong parsing loop being invoked. When you removed the spaces between the numbers, they are interpreted as a single item and therefore the line length is different than the normal systematics line, which meant the parsing did work.
This behaviour is of course not intended. Since the second and third numeric argument passed in the datacard for autoMCStats default to 0 and 1, you can run the validation tool with just ch1 autoMCStats 0, or you can pick up the change from #274, which we can hopefully merge soon.

from combineharvester.

Thank you, @adewit !

from combineharvester.