Package of several technical changes: downgrade of edm::Warnings to edm::Info, removal of an unused parameter from cfi file, upward move of jet selection for tau seeding, switch off of production of null taus #17620
Conversation
…i-mu discriminants for boosted taus
…drons in the same way as jets seeding tau-reco
Fixes for 90X
…u-pog/cmssw into CMSSW_9_0_X_tau-pog_LogWarningFix
|
A new Pull Request was created by @roger-wolf (Roger Wolf) for CMSSW_9_0_X. It involves the following packages: CommonTools/ParticleFlow @cmsbuild, @cvuosalo, @slava77, @monttj, @davidlange6 can you please review it and eventually sign? Thanks. cms-bot commands are listed here #13028 |
|
@cmsbuild please test |
|
The tests are being triggered in jenkins. |
|
+1 The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic: |
|
Comparison job queued. |
|
@roger-wolf |
roger-wolf
changed the title
|
Hi Slava, I hope this satisfies your justified formal request. Comments/confirmations on the more relevant physics question that you are having I've to leave to Michal. Cheers, |
|
Hi Slava, probably what causes confusion is the name "null taus"; Null taus are not taus with zero momentum, but taus build from jets where nothing tau-like was found. Such a tau carries 4-momentum of original seeding jet (pt>14 & |eta|<2.4 by construction) while other data members of PFTau object are not filled thus tau is null (for instance does not possess signal candidates). Null taus were produced for debugging (in this way there is 1-to-1 correspondence between seeding jets and PFtaus) and should be switched off long time ago. Anyway, such taus are not used by analysts and not stored in MiniAOD. There is also an another class of taus which are not quite tau-like. They have an reconstructed decay mode, but it is very poor, e.g. a signal charged hadron candidate for unique tau "prong" is substituted by neutral candidate matched to a (poor) general track. Such taus have usually low or even undefined (null) momentum. Those should be also eliminated, but it requires an additional selection (at least one signal charged hadron to be required). And again, such taus are not stored in MiniAOD. So, propose to revert buildNullTaus to True in this PR and prepare a new PR with set it to false and an additional selection to kill the other type of bad taus. Alternatively, the additional section can be added here, if it is your preference. |
|
On 2/27/17 1:10 AM, Roger Wolf wrote:
Hi Slava,
I hope this satisfies your justified formal request.
Hi Roger,
This is descriptive now.
Thank you.
…--slava
Comments/confirmations on the more relevant physics question that you
are having I've to leave to Michal.
Cheers,
Roger
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#17620 (comment)>, or
mute the thread
<https://github.com/notifications/unsubscribe-auth/AEdcbhwz_4OaQX-P29rhA13ziWQsWHB0ks5rgpL3gaJpZM4MKNMB>.
|
|
On 2/27/17 4:31 AM, mbluj wrote:
So, propose to revert buildNullTaus to True in this PR and prepare a new
PR with set it to false and an additional selection to kill the other
type of bad taus. Alternatively, the additional section can be added
here, if it is your preference.
Fine with me either way.
Maybe indeed better to have this much less technical change (changes
event content) in a separate PR.
|
|
buildNullTaus reverted to previous setting (True) in RecoTauTag/RecoTau/python/RecoTauCombinatoricProducer_cfi.py |
|
@cmsbuild please test |
|
The tests are being triggered in jenkins. |
|
Comparison job queued. |
|
+1 for #17620 8dd7bd3
|
Hi,
I am citing Michal's comment to this pull request below (*). Matrix tests have been performed and were positive. This is pure technical work that should not touch the physics performance. This is also what our tests indicated. Runtime performance might change e.g. since edm::Warnings are dowgrade to edm::Infos.
Cheers,
Roger
(*)
The following issues are fixed with this PR: