Adding plots vs TPC occupancy

[chiarazampolli]

No description provided.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0

[chiarazampolli]

Do not review yet, I need to see if it compiles with QC.

[chiarazampolli]

Hello @shahor02 ,

If it compiles fine, this would be ready for review.

Chiara

[chiarazampolli]

An example from 3 TFs from PbPb (before fixing the Z axis title which is in the last commit):

And from pp:

I am not sure why the multiplicity shows an underflow (in pp; in PbPb there are 15)... The limit is 0, so I would assume there is no underflow.

[chiarazampolli]

As for the underflow, can it be some artifact? If I take the projectionZ itself, I see that the bin with 15 entries is not the underflow one:

{code}
root [3] .x plot.C
ibinZ = 0 low edge = -1e+06 content = 0
ibinZ = 1 low edge = 0 content = 15
ibinZ = 2 low edge = 1e+06 content = 466
ibinZ = 3 low edge = 2e+06 content = 1148
ibinZ = 4 low edge = 3e+06 content = 1139
ibinZ = 5 low edge = 4e+06 content = 951
ibinZ = 6 low edge = 5e+06 content = 2816
{code}

The macro I used to plot and then print the bins is:

{code}
void plot(bool isPbPb = true) {

TFile* f = new TFile("ITSTPCmatched_allTracks_enableK0.root");
TString plotName = "mK0MassVsPtVsOccPbPb";
if (isPbPb == false) {
plotName = "mK0MassVsPtVsOccpp";
}
TH3F* h1 = (TH3F*)f->Get(Form("%s", plotName.Data()));
TH1D* hx = (TH1D*)h1->ProjectionX();
TH1D* hy = (TH1D*)h1->ProjectionY();
TH1D* hz = (TH1D*)h1->ProjectionZ();
gStyle->SetOptStat(1111111);

TCanvas* c = new TCanvas("c", "c", 50, 50, 1150, 1150);
c->Divide(2, 2);
c->cd(1);
h1->Draw();
c->cd(2);
hx->Draw();
c->cd(3);
hy->Draw();
c->cd(4);
hz->Draw();

for (int ibinX = 1; ibinX <= h1->GetNbinsX(); ++ibinX) {
for (int ibinY = 1; ibinY <= h1->GetNbinsY(); ++ibinY) {
if (h1->GetBinContent(ibinX, ibinY, 0) != 0) {
std::cout << "ibinX = " << ibinX << " ibiny = " << ibinY << " content with binZ=0 = " << h1->GetBinContent(ibinX, ibinY, 0) << std::endl;
}
}
}
for (int ibinZ = 0; ibinZ <= hz->GetNbinsX(); ++ibinZ) {
std::cout << "ibinZ = " << ibinZ << " low edge = " << hz->GetXaxis()->GetBinLowEdge(ibinZ) << " content = " << hz->GetBinContent(ibinZ) << std::endl;
}
// c->SaveAs("K0s.png");
//c->SaveAs("K0s.pdf");
return;
}
{code}

[chiarazampolli]

The observation about underflow is simply caused by the fact that I have changed to start from occupancy 0 but probably did not re-run all tests (I am doing 4, pp, PbPb, with and without QC each). Re-running everything, all is good, and when the occ axis starts from 0, there are no underflows.

pp:

with:
ibinZ = 0 low edge = -166667 content = 0
ibinZ = 1 low edge = 0 content = 1
ibinZ = 2 low edge = 166667 content = 2
ibinZ = 3 low edge = 333333 content = 37
ibinZ = 4 low edge = 500000 content = 164
ibinZ = 5 low edge = 666667 content = 35
ibinZ = 6 low edge = 833333 content = 2

PbPb:

with:

ibinZ = 0 low edge = -1e+06 content = 0
ibinZ = 1 low edge = 0 content = 15
ibinZ = 2 low edge = 1e+06 content = 466
ibinZ = 3 low edge = 2e+06 content = 1148
ibinZ = 4 low edge = 3e+06 content = 1139
ibinZ = 5 low edge = 4e+06 content = 951
ibinZ = 6 low edge = 5e+06 content = 2816

[chiarazampolli]

Moving to WIP. The current binning in PbPb, with the reco of a full CTF at ~ 26 kHz shows 1.7% of overflow in TPC occupancy. I am testing with a wider range.

[chiarazampolli]

Hello @shahor02 ,

For me this can be merged.
You will see that there are 2 histograms 3D, one for pp and one for PbPb, where the binning in TPC occupancy is different. Within the QC framework, one could decide which one to use, but it needs to be checked and implemented in the async merging stage.
Without the QC framework, one could then use the configParam.
For now, we have two histograms, one will be empty.

Chiara

[chiarazampolli]

With a second thought, maybe it is better to add the cuts on DCA and cosPA before merging...

[shahor02]

OK, I would make the cuts the parameters of the json.
To avoid defining 2 histos: cannot we use different jsons (passing the binning) depending on the $BEAMTYPE in the qc-workflow.sh ?

[chiarazampolli]

Yes, sure, the cuts will be a parameter.
And yes, 2 jsons were the first thought I had, then I found this:
https://github.com/AliceO2Group/QualityControl/blob/master/doc/Advanced.md#definition-and-access-of-user-defined-configuration-extendedtaskparameters
which in online would work very well, but for async we need to see for the merging, it seems.
I think that maintaining 2 json is very error prone, this is why I would not do it unless this extra histogram is really a problem.

[chiarazampolli]

With the cut on DCA and cosPA, 1 full CTF from zzo (PbPb): statistics decrease from ~170k to ~3k candidates:

[alibuild]

Error while checking build/O2/fullCI for b807484 at 2024-10-10 14:03:

## sw/BUILD/O2-latest/log
c++: error: unrecognized command-line option '--rtlib=compiler-rt'
c++: error: unrecognized command-line option '--rtlib=compiler-rt'
input_line_9:367:51: error: cannot assign to non-static data member within const member function 'setK0MinCosPA'
Error: /sw/slc8_x86-64/ROOT/v6-32-06-alice1-7/bin/rootcling: compilation failure (/sw/BUILD/ecff9fc079851887d3c56be5995620165734b40e/O2/Detectors/GLOQC/G__O2GLOQCfb38b73dd3_dictUmbrella.h)
ninja: build stopped: subcommand failed.

Full log here.

[chiarazampolli]

Hello @shahor02 ,
Can you review?
Thanks!
Chiara