Integration of the E\gamma Cluster Position Calculator into the PFClusterProducer

[lgray]

@jbendavid @ekfriis @matteosan1 @cms-jet @cms-met

This PR incorporates the result of a study done that demonstrated improved cluster position resolution when the Particle Flow default electron/photon position corrections were replaced with the default E\gamma corrections.

The main improvement is seen in the ECAL endcaps, but there is some slight improvement for the barrel.

You can see the study here:
https://indico.cern.ch/getFile.py/access?contribId=0&resId=0&materialId=slides&confId=270535

I've @'d the Tau POG and jet/met since this should impact their particle-based reconstruction.

[cmsbuild]

A new Pull Request was created by @lgray (Lindsey Gray) for CMSSW_7_0_X.

Integration of the E\gamma Cluster Position Calculator into the PFClusterProducer

It involves the following packages:

RecoEcal/EgammaCoreTools
RecoParticleFlow/Configuration
RecoParticleFlow/PFClusterProducer

@thspeer, @slava77 can you please review it and eventually sign? Thanks.
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.

[nclopezo]

Hi,

I ran the tests on top of CMSSW_7_0_X_2013-09-04-0200, all tests passed:

https://cmssdt.cern.ch/jenkins/job/Pull-Request-Integration/ARCHITECTURE=slc5_amd64_gcc472/381/console

you can see the artifacts here:

https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/381

[slava77]

working on it @slava77

[slava77]

I think this should be a cms::Exception

[slava77]

@lgray
Lindsey, please follow up on timing performance of particleFlowClusterECAL
The new version is taking ~4-6 times more depending on the test workflow
Thanks

[slava77]

+1

(If this is still making it to pre4, please merge, otherwise, we can take a bit of time to solve CPU/timing degradation)

tested a311925 in CMSSW_7_0_X_2013-09-10-0200

As expected, observe significant improvement (mostly in endcaps) of reco-true matching in phi (mainly) and a bit in eta.
This is visible in electron and photon gun samples, nothing is really visible in high-PU or QCD samples, as it should be (no real stuff to do matching here).

I observe no degradation in MET.
Supposedly, there should also be an improvement in tau reco (mostly a guess, nothing visible in the test samples, which don't really have clean taus)

Main regression is in CPU performance: increase of a factor of 4-6 in timing of particleFlowClusterECAL.
Physics improvements would benefit more from a full validation cycle now.

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next IBs unless changes or unless it breaks tests. @ktf can you please take care of it?