-
Notifications
You must be signed in to change notification settings - Fork 4.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Disabled PCA inputs for deepTau v2 (10_6_X) #27882
Disabled PCA inputs for deepTau v2 (10_6_X) #27882
Conversation
A new Pull Request was created by @mbluj for CMSSW_10_6_X. It involves the following packages: RecoTauTag/RecoTau @perrotta, @cmsbuild, @slava77 can you please review it and eventually sign? Thanks. cms-bot commands are listed here
|
@cmsbuild please test |
The tests are being triggered in jenkins. |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
changes to physics are nominally not allowed in a production release (no-change policy). |
In this case, is it useful to already migrate this backport with the preliminary modifier? Or should we rather keep the PR open until the proper modifier is available? |
|
{closed by mistake, so reopened, sorry) |
it may end up in a "case by case" situation. |
I suggest to use the existing modifier now. On our side, we should create an issue to keep track of this and other requests for reminiAOD so that this feature does not get lost. |
I created #27889 |
+operations the update is coherent with the strategy agreed for the backport of this code |
+xpog |
+1 |
merge |
+1 |
This pull request is fully signed and it will be integrated in one of the next CMSSW_10_6_X IBs (tests are also fine) and once validation in the development release cycle CMSSW_11_0_X is complete. This pull request will be automatically merged. |
@peruzzim unfortunately this does not work out of the box:
in at least 3 test workflows 1329.1, 136.772 and 136.8521 . @mbluj I temporarily revert the PR waiting for a fix , so as to move forward with a build for other independent needs |
It is strange as the PR passed matrixTests and the NanoAOD workfolw has been particularly tested by myself... I will take a look ASAP. |
I checked and found that the there are conflicting era modifiers in failing workflows. |
Problem is fixed in the most recent commit by proper ordering era modifiers in the |
Thank you Michal. Should the same fix also get applied to the master version of this PR? |
OK, I am preparing PR to the revert branch with hope that it is what you mean here.
No, it is not needed neither for master nor for 10_2_X; the issue is related only to 10_6_X where new specific tau_ul eras have been added and then chained to master Run201x ones. |
indeed we haven't seen any problem in master, this is due to the adjustment required by backport |
PR description:
Problems have been detected with dxy_PCA input variables to DeepTauID which are inconsistent between data and MC. This PR sets input values to 0 and disables the variables this way.
Issue and fix were discussed in https://indico.cern.ch/event/842796/contributions/3541592/attachments/1897386/3130770/2019-08-26_DeepTau_ID_2017v2_DY_MC_vs_Embedded_and_bug_in_definition_of_PCA_variables.pdf (see slide 7 in particular)
Backport of #27878 to 10_6_X for ultra-legacy MiniAODv2.
PR validation:
compiles & re-MiniAOD test (runTheMatrix.py -l 1325.51 -i all --ibeos)
if this PR is a backport please specify the original PR:
#27878
@swozniewski, FYI