First step towards the new centrality task #122

victor-gonzalez · 2021-09-27T15:38:58Z

So far a separated new task only tested by the
dpt-dpt correlations task

jgrosseo · 2021-09-28T07:37:17Z

Common/TableProducer/centralityTable_new.cxx

+  {
+    /* check the previous run number */
+    auto bc = collision.bc_as<aod::BCsWithTimestamps>();
+    if (bc.runNumber() != mRunNumber) {


This is for caching, right? In the future the CCDBConfigurable will do this automatically.

Good!
For the time being if the run number does not change the CCDB is only queried once

Common/TableProducer/centralityTable_new.cxx

Common/DataModel/Centrality_new.h

victor-gonzalez · 2021-09-28T08:16:13Z

Sorry!
Wrong action!
Not ready for reviewing yet!
Still in draft!

victor-gonzalez · 2021-09-28T08:51:28Z

@ekryshen I don't know how to formally add you to the reviewers of this draft PR
I hope it will be enough with this
Your inputs will be really appreciated
Thanks in advance!

ddobrigk · 2021-10-05T11:29:38Z

Hi @victor-gonzalez, thank you very very much, this looks very good! Note that the files I sent you contain also other estimators, such as CL0 and CL1, in them already and that this could be done in a second step.

For posterity, we may also want to have a broader discussion (with @jgrosseo and others) on standardizing estimator naming and especially how to proceed with the actual Run 3 data, where we may want to use FIT signals etc.

victor-gonzalez · 2021-10-05T12:44:49Z

Thanks @ddobrigk
Yes, the information in CCDB for CL0, CL1 and the other estimators, is already normalized and ready to be used by the centrality task. As the only information which is currently available/required is based on the V0M estimator, I believe it is better to go to production with only that estimator available (although based in V0A and V0C) and then progressively modify the centrality task to incorporate the other ones

jgrosseo · 2021-10-05T13:14:50Z

Yes, before merging we should discuss the column names. Some of them should include Run2 I guess because we will not have an SPD in Run 3 for instance :)

victor-gonzalez · 2021-10-10T18:40:02Z

Now the PR incorporates the side effects on all affected files
The checks run correctly with three data files from LHC15o two of the same run and one additional one from a different run
The previous centrality task is replaced by the new one within the original source files

Common/DataModel/Centrality.h

jgrosseo · 2021-10-11T13:23:06Z

Very nice! We should add your converted centrality files then to the production CCDB - I have asked access for you. What would be a good prefix? Analysis/Centrality ?

victor-gonzalez · 2021-10-11T13:29:01Z

Analysis/Centrality seems appropriate
I will update the PR

ekryshen · 2021-10-11T19:10:46Z

Very nice! We should add your converted centrality files then to the production CCDB - I have asked access for you. What would be a good prefix? Analysis/Centrality ?

We already have /Centrality folder in the production CCDB:
http://alice-ccdb.cern.ch/browse
Similar to /EventSelection
Do you think it would be better to move everything to /Analysis? Then we should ask Costin to clean up.

jgrosseo · 2021-10-12T06:16:47Z

Very nice! We should add your converted centrality files then to the production CCDB - I have asked access for you. What would be a good prefix? Analysis/Centrality ?

We already have /Centrality folder in the production CCDB: http://alice-ccdb.cern.ch/browse Similar to /EventSelection Do you think it would be better to move everything to /Analysis? Then we should ask Costin to clean up.

Sorry for this. Yes, we should keep the old path. However, then we cannot upload yet the objects, because the old centrality task will stop working...

jgrosseo · 2021-10-12T06:17:54Z

@victor-gonzalez I would say then you adjust the table names (as discussed above), resolve the conflicts, and we ask people to test this PR (still using the test-ccdb). Then we can make a upload & merge in the same moment.

victor-gonzalez · 2021-10-12T08:41:29Z

Very good!
The upload takes few seconds only
Should we leave the current calibration for LHC15o?
I can remove it before uploading the new one

victor-gonzalez · 2021-10-13T11:06:10Z

Very nice! We should add your converted centrality files then to the production CCDB - I have asked access for you. What would be a good prefix? Analysis/Centrality ?

We already have /Centrality folder in the production CCDB: http://alice-ccdb.cern.ch/browse Similar to /EventSelection Do you think it would be better to move everything to /Analysis? Then we should ask Costin to clean up.

Sorry for this. Yes, we should keep the old path. However, then we cannot upload yet the objects, because the old centrality task will stop working...

The old path is Centrality/CumMultV0M which is not any longer representative because the new calibration information incorporates estimation from other detectors.
I guess we should use directly the Centrality prefix
But having in mind that the next step we are addressing is the pp multiplicity percentiles, does it make sense to have a common CentralityMultiplicity prefix?
@ddobrigk What do you think?

jgrosseo · 2021-10-13T11:08:58Z

Thanks. So we will not have an issue with name clashes.
I see your point but CentralityMultiplicity is not nice. So due to lack of other suggestions I'd keep Centrality
You can then already upload your objects into Centrality/Estimators ?

victor-gonzalez · 2021-10-13T11:56:33Z

Tried, but I can only do it at http://alice-ccdb.cern.ch:8080/Centrality/Estimators
If I do it at http://alice-ccdb.cern.ch/Centrality/Estimators I got rejected
If done at http://alice-ccdb.cern.ch:8080 will it be automatically transferred to http://alice-ccdb.cern.ch at some point?

victor-gonzalez · 2021-10-13T12:42:18Z

Also the conflicts were resolved

victor-gonzalez · 2021-10-13T12:45:47Z

I'm not sure I understand the clang-format check error

victor-gonzalez · 2021-10-13T13:06:52Z

Ok!
Now I see them

victor-gonzalez · 2021-10-14T14:56:25Z

LHC15o, LHC18q and LHC18r centrality calibration and estimation information is already uploaded to the CCDB production endpoint under Centrality/Estimators
Nice output from the centrality-qa from three data files from two different runs from LHC15o
https://cernbox.cern.ch/index.php/s/PBnkasvlHFxkJgO

jgrosseo · 2021-10-15T06:51:09Z

It does not compile. Can you rebase and update?

So far a separated new task only tested by the dpt-dpt correlations task

Also minor typos

Based on V0M estimation but ready for other estimators implementation

* Separate functions to alleviate memory issues * Please consider the following formatting changes (#122) --------- Co-authored-by: David Dobrigkeit Chinellato <david.dobrigkeit.chinellato.cern.ch> Co-authored-by: ALICE Builder <alibuild@users.noreply.github.com>

* Separate functions to alleviate memory issues * Please consider the following formatting changes (AliceO2Group#122) --------- Co-authored-by: David Dobrigkeit Chinellato <david.dobrigkeit.chinellato.cern.ch> Co-authored-by: ALICE Builder <alibuild@users.noreply.github.com>

[PWGLF] Update spectraTOF.cxx