implementation of TuneOnData for MC in TPC PID

[jezwilkinson]

Original pull request for TuneOnData #4733 had diverged due to large conflicting changes made in #5563. This commit resolves the conflicts and re-implements the tuneondata option based on the new NN configurables (beta-gamma cutoff and per-species "useNetwork" flags).

[jezwilkinson]

(for information: @wiechula@njacazio )

[alibuild]

Error while checking build/O2Physics/o2 for 7a30be3 at 2024-04-16 13:16:

## sw/BUILD/O2Physics-latest/log
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPCFull.cxx:74:21: error: 'mcTPCTuneOnData' is not a member of 'o2::aod'
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPCFull.cxx:74:36: error: template argument 1 is invalid
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPCFull.cxx:486:21: error: request for member 'reserve' in '((tpcPidFull*)this)->tpcPidFull::tableTuneOnData', which is of non-class type 'int'
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPCFull.cxx:549:22: error: expression cannot be used as a function
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPCFull.cxx:598:37: error: 'class o2::pid::tpc::Response' has no member named 'GetNumberOfSigmaMCTuned'; did you mean 'GetNumberOfSigma'?
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPC.cxx:74:21: error: 'mcTPCTuneOnData' is not a member of 'o2::aod'
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPC.cxx:74:36: error: template argument 1 is invalid
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPC.cxx:480:21: error: request for member 'reserve' in '((tpcPid*)this)->tpcPid::tableTuneOnData', which is of non-class type 'int'
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPC.cxx:542:22: error: expression cannot be used as a function
/sw/SOURCES/O2Physics/5633-slc7_x86-64/0/Common/TableProducer/PID/pidTPC.cxx:588:75: error: 'class o2::pid::tpc::Response' has no member named 'GetNumberOfSigmaMCTuned'; did you mean 'GetNumberOfSigma'?
ninja: build stopped: subcommand failed.

Full log here.

[njacazio]

Hi, this looks good, shall we release it and enable a test wagon so that analysers can test it?

[wiechula]

Fine from my side. But I would not enable it by default.

[njacazio]

Fine from my side. But I would not enable it by default.

I fully agree

[njacazio]

Hi @jezwilkinson what do you think? Can we go ahead?

[jezwilkinson]

Hi @njacazio ,
Sorry, I was away on a group event and completely offline the last couple of days.
Sounds great to me!

"I would not enable it by default."

Indeed the process switch is defaulted to false (not least because the TuneOnData process loop requires subscription to MC info and so wouldn't work on real data). We can announce in the mattermost how to use it

[ddobrigk]

thanks a lot!

[njacazio]

Hi, I propose that we add a debug wagon with the tune on data enabled. I will do this tomorrow with the new tag and test it

[jgrosseo]

@jezwilkinson@ddobrigk This PR introduced a problem: Now TPC full and TPC tiny produce both the same table which means on Hyperloop wagons which require full and wagons which require tiny cannot be combined into a train anymore. It should be removed from one of the two, and if that is not good, be moved to a separate executable...

[njacazio]

Indeed, for now we could split the two tables #5946