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
Fix for JPTFastjet correction chains #18121
Conversation
A new Pull Request was created by @kodolova (kodolova) for master. It involves the following packages: JetMETCorrections/Configuration @perrotta, @cmsbuild, @slava77, @monttj, @davidlange6 can you please review it and eventually sign? Thanks. cms-bot commands are listed here #13028 |
please test |
The tests are being triggered in jenkins. |
Comparison job queued. |
@kodolova : |
This is the correct implementation of the JPT jets. It takes both L1Calo and ZSP corrections and make a combination. Chains should start with call for L1JPTOffsetCorrectorProducer |
Henning Kirschenmann hmm, I think the change that is in conflict with your PR is a huge one by David Dagenhart, touching "almost everything" in CMSSW: Definitely nothing JetMET-specific. Obviously, nobody from JetMET is trying to revert your changes in any way. Sorry for the extra-trouble, but I think you have to rebase your changes on the latest branch. Thanks for maintaining/bugfixing those JPT-sequences in any case! Cheers, |
No need to send any copy of the message! I only meant that I got confused by the comment in this PR, which was referred to the old one instead, and that's why I did not start reviewing it at once. (Too late, you already sent it... ok, thank you anyhow) About the fix, I'm not questioning its validty or correctness! I only would like to see the effect of that fix on at least one workflow. You say that there is a 20% effect on JES: how can I see (on validation plots, DQM or whatever) such a 20% modification of the jet energy before and after applying this fix to the baseline release? |
I do not have plots. I can do if needed. Then with current configuration I get: With new tag I get: The difference depends on pt jet, as ZSP corrections is the correction for the Calopart of jet and depends on Calojet pt. |
Thank you, that's fine with me. |
I changed Ak4 to AK5 for Calojets. I checked with runMatrix - no errors. |
please test |
The tests are being triggered in jenkins. |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
+1 Of course, this is just to have the L1Fastjet chain is in place for JPT; as soon as the AK4 payloads for calojets will be available in the relevant GTs the authors will take care of updating the configurations touched by this PR |
cross-post from private email thread Dear Olga, we looked further into the details to find out that the error from a missing record+label (label: AK4Calo, AK4JPT ) pair emerged in the run1 8 TeV workflow. Such labels are present in the 2017 data and MC workflows. Why do you need to modify the jet correction sequence of run1 ? I would imagine you woulds handle such sequence by era, and focus on the 13 TeV scenarios. . do the relevant tags measured at 8 TeV exist ? The tags with the same label present in the 13 TeV GT's are not necessarily consistent with usage at 8 TeV . if they exist, the tags need to be presented, queued and announced in hn, see the procedure documented here https://twiki.cern.ch/twiki/bin/view/CMS/UserTagsInTheGTProcedure#Current_procedure_during_Run_II HTH, Cheers, |
Answer to private mail thread Dear Giovanni,
Ideally for 8 TeV it should be ak5 reconstruction chain in Validation tests and for 13 TeV ak4 reconstruction chain. |
On 4/13/17 4:09 AM, kodolova wrote:
Ideally for 8 TeV it should be ak5 reconstruction chain in Validation
tests and for 13 TeV ak4 reconstruction chain.
this is not what we do in CMSSW processing.
Everything in a given release is supposed to be processed with the same
latest configuration
and eras and GTs are meant to address only detector-specific changes.
Eras and GTs are not meant to create historical algorithmic choices.
If we have to support AK5 in 91X for JPT, it will have to be for all
MC/data ranges or eras.
|
We do not need to support AK5 for 91X for JPT. The problem is not with JPT. The problem is with Calojets and the historical artefact of jet energy corrections configuration. |
@davidlange6 : the last discussion was on the part of this PR which got reverted (i.e. ak4 for calojets) , and therefore kept out of this PR. It must continue, but probably it is better if it is out of this thread. For what this PR is concerned, my comment in #18121 (comment) still stands: "Of course, this is just to have the L1Fastjet chain is in place for JPT; as soon as the AK4 payloads for calojets will be available in the relevant GTs the authors will take care of updating the configurations touched by this PR" |
I discussed the new GT with JERC subgroup of JetMET. They promised to have it within 2 weeks but they promised to have new corrections already a month ago. These 2 weeks may come to month or so. May I propose to integrate this PR as it is if all tests as ok. |
Hi @perrotta - i'm not sure if I can determine a "yes" or "no" from your answer. if the PR is not ready to go, could you put it on hold, otherwise I'll assume your +1 still stands |
@davidlange6 : my answer meant "yes, I confirm my +1 (and I assume that further updates will come in a future new PR)" |
This request follows PR17900 for changes with Task
The reason of change is that JPTjets have the L1 corrections in two steps: firstly Calojets (part of JPTjet) are corrected for L1Offset or L1Fastjet and then corrections for zero suppression is done. Currently it is correctily implemented only for L1Offset. For chains that includes CaloL1Fastjet, zsp corrections are missed and this is 20% of jet energy scale.
Concerning rebase/update, I see that Configuration is completely changed. Thus, this is not a matter of rebase. I sent message to JetMET conveners and ask how to proceed.