Fix crash in pat::MET when no Type-I uncertainties are added

[blinkseb]

Hello,

This PR fixes a crash I experienced when running on CMSSW 7.4.12 patch4. I manually create a new slimmedMETs using the following code snippet:

from JMEAnalysis.JetToolbox.jetToolbox_cff import jetToolbox
jetToolbox(process, 'ak4', 'ak4CHSJetSequence', 'out', PUMethod='CHS', runOnMC=not isData, miniAOD=True, addPUJetID=False)

from PhysicsTools.PatAlgos.tools.metTools import addMETCollection
process.load('RecoMET.METProducers.PFMET_cfi')

process.pfMet.src = cms.InputTag('packedPFCandidates')
addMETCollection(process, labelName='patPFMet', metSource='pfMet')


process.load('JetMETCorrections.Configuration.JetCorrectors_cff')
from JetMETCorrections.Type1MET.correctionTermsPfMetType1Type2_cff import corrPfMetType1
from JetMETCorrections.Type1MET.correctedMet_cff import pfMetT1

process.corrPfMetType1 = corrPfMetType1.clone(
        src = 'ak4PFJetsCHS',
        jetCorrLabel = 'ak4PFCHSL1FastL2L3Corrector' if not isData else 'ak4PFCHSL1FastL2L3ResidualCorrector',
        offsetCorrLabel = 'ak4PFCHSL1FastjetCorrector'
    )
    process.pfMetT1 = pfMetT1.clone(
        src = 'pfMet',
        srcCorrections = [cms.InputTag("corrPfMetType1", "type1")]
    )

addMETCollection(process, labelName='patMET', metSource='pfMetT1') # T1 MET

from PhysicsTools.PatAlgos.slimming.slimmedMETs_cfi import slimmedMETs
del slimmedMETs.caloMET

process.slimmedMETs = slimmedMETs.clone()

process.slimmedMETs.src = cms.InputTag("patMET")
process.slimmedMETs.rawVariation = cms.InputTag("patPFMet")

I explicitly do not want T1 uncertainties, I'm only interested in T1 met + Raw met.

When trying to access pat::MET::uncorPt() the code crash inside the pat::MET::findMETTotalShift.

This is because in this particular case, the uncertainties_ array is empty, because, since there's no Type1 uncertainties, the function setUncShift is never called by PATMETSlimmer.

This PR fixes both issues: first, it ensure that if uncertainties_ is empty, the code no longer crashed. Second, it fixes PATMETSlimmer by making sure setUncShift is called not matter the type of uncertainties.

This PR changes pat::MET::findMETTotalShift to prevent the crash by checking if uncertainties_ is empty.

If merged, it'll need to be backport to 7.5 and 7.4.

Thanks

[cmsbuild]

A new Pull Request was created by @blinkseb (Sébastien Brochet) for CMSSW_7_6_X.

Fix crash in pat::MET when no Type-I uncertainties are added

It involves the following packages:

DataFormats/PatCandidates
PhysicsTools/PatAlgos

@cmsbuild, @vadler, @monttj can you please review it and eventually sign? Thanks.
@TaiSakuma, @imarches, @ahinzmann, @acaudron, @mmarionncern, @rappoccio, @jdolen, @nhanvtran, @schoef, @ferencek, @gpetruc, @mariadalfonso, @pvmulder this is something you requested to watch as well.
You can sign-off by replying to this message having '+1' in the first line of your reply.
You can reject by replying to this message having '-1' in the first line of your reply.
If you are a L2 or a release manager you can ask for tests by saying 'please test' in the first line of a comment.
@Degano you are the release manager for this.
You can merge this pull request by typing 'merge' in the first line of your comment.

[mmarionncern]

That line should be removed, as it will conflict with the storage of the MET type1, the way it is done, it will erase the type1MET to put the type1Smeared MET in place

[blinkseb]

Ok, I wasn't sure about this one. I'll remove up, thanks!

[blinkseb]

I updated the PR with @mmarionncern comments.

[cmsbuild]

Pull request #11394 was updated. @cmsbuild, @vadler, @monttj can you please check and sign again.

[mmarionncern]

+1 for me

[mariadalfonso]

@blinkseb Can you make a PR in the 74X and 75X we can aim to have it in an analysis release.

[monttj]

please test

[cmsbuild]

The tests are being triggered in jenkins.

[blinkseb]

@mariadalfonso I opened two new PR for 7.4.x ( #11413 ) and 7.5.x ( #11412 )

[lgray]

@blinkseb I only see the changes you mention in the PR description to pat::MET::findMETTotalShift, there are no changes to PATMETSlimmer in your commit history.

[blinkseb]

@lgray See the discussion history with @mmarionncern ; those changes were not correct. I'll edit the PR description

[cmsbuild]

+1
Tested at: 5a75f34
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-11394/8258/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-11394/8258/summary.html

[blinkseb]

Any updates on this?