TPC: Adding workflow to provide pressure and temperature

[matthias-kleiner]

• will be used for scaling of vdrift and gain

Optimizing temperature processing

• move fitting of temperature to definition of temperature struct
• optimize fit procedure by merging last interval in case it is needed

Adding processing of DCS pressure values and storage in CCDB

• first pressure sensor in the cavern serves as the reference
• second pressure sensor in the cavern and third sensor at the surface are normalized to reference pressure sensor

[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
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

[matthias-kleiner]

Hello @wiechula ,
if the output format of the pressure is fine, I can make the upload request to the CCDB. I already put it in the test CCDB for testing (http://ccdb-test.cern.ch:8080/browse/TPC/Calib/Pressure)

[shahor02]

@matthias-kleiner how the TPCPressureTemperatureSpec device will be used? Does it go to the global reconstruction workflow?

[matthias-kleiner]

@matthias-kleiner how the TPCPressureTemperatureSpec device will be used? Does it go to the global reconstruction workflow?

I think it will be used wherever the v-Drift is used. Perhaps @wiechula can comment on this?

[wiechula]

@shahor02 , yes, the idea is to inject the pressure and temperature values into the data stream during reconstruction.
It should then be used in the vdrift and gain calibration.

[wiechula]

@matthias-kleiner , thanks a lot! For me this looks fine.

[shahor02]

@matthias-kleiner @wiechula I am wondering if it is worth running a separate processes just to send downstream a couple of numbers. This might be necessary in case e.g. ITS-TPC matching running with tracks,clusters input from the files, but for full production workflow it would be more economical to call a helper from the TPC reco device and send the values from it.

[wiechula]

@matthias-kleiner , could you implement such a helper? I think we can also keep the separate process.

[matthias-kleiner]

@shahor02 Could you explain how such a helper function should be implemented and where exactly to put it?
@wiechula can correct me, but I think we would also use it in the tpc digitizer to simulate the dependence of gain and vdrift on p/T and use it not only in the reconstruction.

[shahor02]

@matthias-kleiner my main concern is adding a new process which will cost ~100 MB or more just to send a few numbers.
I could be a helper like the TPC VDriftHelper, which knows how to add needed CCDB inputs to the externally provided inputs vector and how to digest them, see

AliceO2/Detectors/TPC/calibration/src/VDriftHelper.cxx

Lines 127 to 203 in 0c08a1f

	void VDriftHelper::extractCCDBInputs(ProcessingContext& pc, bool laser, bool itstpcTgl)
	{
	if (mForceParamDrift && mForceParamOffset) { // fixed from the command line
	return;
	}
	if (laser && !mForceParamDrift) {
	pc.inputs().get<o2::tpc::LtrCalibData*>("laserCalib");
	}
	if (itstpcTgl) {
	pc.inputs().get<o2::tpc::VDriftCorrFact*>("vdriftTgl");
	}
	if (mUpdated) { // there was a change
	// prefer among laser and tgl VDrift the one with the latest update time
	auto saveVD = mVD;
	mVD = mVDTPCITSTgl.creationTime < mVDLaser.creationTime ? mVDLaser : mVDTPCITSTgl;
	auto& loserVD = mVDTPCITSTgl.creationTime < mVDLaser.creationTime ? mVDTPCITSTgl : mVDLaser;
	if (mForceParamDrift) {
	mVD.refVDrift = saveVD.refVDrift;
	mVD.corrFact = saveVD.corrFact;
	mVD.corrFactErr = 0.f;
	}
	if (mForceParamOffset) {
	mVD.refTimeOffset = saveVD.refTimeOffset;
	mVD.timeOffsetCorr = 0.f;
	}
	mSource = mVDTPCITSTgl.creationTime < mVDLaser.creationTime ? Source::Laser : Source::ITSTPCTgl;
	auto loseCTime = loserVD.creationTime;
	loserVD = mVD; // override alternative VD to avoid normalization problems later
	loserVD.creationTime = loseCTime;
	std::string rep = fmt::format("Prefer TPC Drift from {} with time {} to {} with time {}",
	SourceNames[int(mSource)], mVD.creationTime, mSource == Source::Laser ? SourceNames[int(Source::ITSTPCTgl)] : SourceNames[int(Source::Laser)],
	mSource == Source::Laser ? mVDTPCITSTgl.creationTime : mVDLaser.creationTime);
	if (mForceParamDrift || mForceParamOffset) {
	std::string impos = mForceParamDrift ? "VDrift" : "";
	if (mForceParamOffset) {
	impos += mForceParamDrift ? " and DriftTimeOffset" : "DriftTimeOffset";
	}
	rep += fmt::format(" but {} imposed from command line", impos);
	}
	LOGP(info, "{}", rep);
	}
	}
	//________________________________________________________
	void VDriftHelper::requestCCDBInputs(std::vector<InputSpec>& inputs, bool laser, bool itstpcTgl)
	{
	if (laser) {
	addInput(inputs, {"laserCalib", "TPC", "CalibLaserTracks", 0, Lifetime::Condition, ccdbParamSpec(CDBTypeMap.at(CDBType::CalLaserTracks))});
	}
	if (itstpcTgl) {
	// VDrift calibration may change during the run (in opposite to Laser calibration, at least at the moment), so ask per-TF query
	addInput(inputs, {"vdriftTgl", "TPC", "VDriftTgl", 0, Lifetime::Condition, ccdbParamSpec(CDBTypeMap.at(CDBType::CalVDriftTgl), {}, 1)});
	}
	}
	//________________________________________________________
	void VDriftHelper::addInput(std::vector<InputSpec>& inputs, InputSpec&& isp)
	{
	if (std::find(inputs.begin(), inputs.end(), isp) == inputs.end()) {
	inputs.emplace_back(isp);
	}
	}
	//________________________________________________________
	bool VDriftHelper::accountCCDBInputs(const ConcreteDataMatcher& matcher, void* obj)
	{
	if (matcher == ConcreteDataMatcher("TPC", "VDriftTgl", 0)) {
	accountDriftCorrectionITSTPCTgl(static_cast<VDriftCorrFact*>(obj));
	return true;
	}
	if (matcher == ConcreteDataMatcher("TPC", "CalibLaserTracks", 0)) {
	accountLaserCalibration(static_cast<LtrCalibData*>(obj));
	return true;
	}
	return false;
	}

Then it can be used either inside the existing device, e.g., as shown in

AliceO2/GPU/Workflow/src/GPUWorkflowTPC.cxx

Line 344 in 0c08a1f

mTPCVDriftHelper->extractCCDBInputs(pc);

, with its output sent out from the same device, or it can be wrapped into a standalone process.

[matthias-kleiner]

Many thanks for the explanation @shahor02 . This makes indeed more sense. I will then implement it as you suggested

[matthias-kleiner]

Hello @shahor02 @wiechula
I implemented the helper function. Is it what you had in mind?

[shahor02]

Exactly, thanks!