Reweighting only works first time

[tom ingebretsen carlson]

I generate an unweighted sample and then I use python ./bin/madevent reweight run_01, it works the first time. Then I change the reweight_card.dat and I do ./bin/madevent reweight run_01 again and it crashes.

Seemingly the problem is that I can't apply the MG reweighting to the same sample twice, as even if I don't change the reweight_card.dat it crashes if I do ./bin/madevent reweight run_01 twice.

[Olivier Mattelaer (olivier-mattelaer)]

Sorry, they are one file missing in your model:
restrict_NLO_rew_cpg.dat

Can you add it?

Cheers,

Olivier

[tom ingebretsen carlson (makronymen)]

No problem. Let me know if you need anything else.

Cheers,

Tom

[Olivier Mattelaer (olivier-mattelaer)]

Sorry, here I do not reproduce the error.

So I did the following run
1) in a file I set the following command
import model SMEFTatNLO_v3-NLO_rew_cpg
define p = 21 2 4 1 3 -2 -4 -1 -3 5 -5 # pass to 5 flavors
define j = p
generate p p > h h QCD = 0 QED = 2
output TEST_RWGT
launch
reweight=ON
./param_card.dat # local copy of the file that you put here
./run_card.dat # local copy of the file that you put here
./reweight_card.dat # local copy of the file that you put here but removing the line specific to cluster.

2) go to TEST_RWGT directory and run
./bin/madevent reweight run_01

and everything was fine.

Cheers,

Olivier

[tom ingebretsen carlson (makronymen)]

Hi,

thanks for the help and the tips.

The problem was when I did the step 2) a second time and I used another process to reweight to compared to the first reweighting, then it crashed.

However I managed to solve it now by removing the rw_me folder. Then it works.

All best,

Tom

[Olivier Mattelaer (olivier-mattelaer)]

I also try with different reweight_card ,...
but always fail to reproduce this...

But since you found a work around I will close this ticket for the moment, if you face this again do not hesitate to comment and I can do more test.

Olivier

[tom ingebretsen carlson (makronymen)]

Okay, thanks!

All best,

Tom