TrigInDetTrackFitter magnetic field migration - ATR-21310

Migrate TrigInDetTrackFitter to new magnetic field cache. Also, TrigL2ResidualCalculator shared most of its code with TrigInDetTrackFitter - add the getUnbiasedResiduals method to TrigInDetTrackFitter and delete the tool.

Mysteriously in my local tests this led to muCombMT complaining about being unable to retrieve its field service:

muCombMT_Muon_FTF                              0   ERROR Could not retrieve AtlasFieldSvc --> Abort
muCombMT_Muon_FTF                              0   ERROR AthenaBaseComps/AthenaBaseComps/AthCommonDataStore.icc:43 (StatusCode AthCommonDataStore<PBASE>::sysInitialize() [with PBASE = AthCommonMsg<Algorithm>]):
l2muCombViewNode                               0   ERROR Unable to initialize Algorithm muCombMT/muCombMT_Muon_FTF

I don't know what's going on here so decided to submit the MR anyway. Maybe muCombMT is somehow relying on the AtlasFieldSvc initialised by TrigInDetTrackFitter up to this point?

Cheers,

Stewart

This merge request affects 3 packages:

• Trigger/TrigAlgorithms/TrigFastTrackFinder
• Trigger/TrigTools/TrigInDetToolInterfaces
• Trigger/TrigTools/TrigInDetTrackFitter

Adding @sutt ,@nagano ,@bernius as watchers

added Trigger master review-pending-level-1 labels

CI Result FAILURE (hash 8cd3338d)

	Athena	AthSimulation	AnalysisBase	AthGeneration
externals
cmake
make
required tests
optional tests

Full details available on this CI monitor view
Athena: number of compilation errors 1, warnings 0
AthSimulation: number of compilation errors 0, warnings 0
AnalysisBase: number of compilation errors 0, warnings 0
AthGeneration: number of compilation errors 0, warnings 0
For experts only: Jenkins output [CI-MERGE-REQUEST-CC7 17347]

/var/lib/jenkins/workspace/CI-MERGE-REQUEST-CC7/master/Trigger/TrigAlgorithms/TrigFastTrackFinder/src/TrigFastTrackFinder.cxx:57:10: fatal error: TrigInDetToolInterfaces/ITrigL2ResidualCalculator.h: No such file or directory
 #include "TrigInDetToolInterfaces/ITrigL2ResidualCalculator.h"

I guess a stray include after the removal?

added review-user-action-required label and removed review-pending-level-1 label

added 1 commit

• de8fc090 - Remove include/forward declaration of ITrigL2ResidualCalculator

Compare with previous version

This merge request affects 3 packages:

• Trigger/TrigAlgorithms/TrigFastTrackFinder
• Trigger/TrigTools/TrigInDetToolInterfaces
• Trigger/TrigTools/TrigInDetTrackFitter

Adding @sutt ,@nagano ,@bernius as watchers

added review-pending-level-1 label and removed review-user-action-required label

Thanks Christos, indeed I missed that.

For the possible muCombMT_Muon_FTF issue you mention I see !34872 (merged) also on going so we shall see ...

CI Result FAILURE (hash de8fc090)

	Athena	AthSimulation	AnalysisBase	AthGeneration
externals
cmake
make
required tests
optional tests

Full details available on this CI monitor view
Athena: number of compilation errors 0, warnings 0
AthSimulation: number of compilation errors 0, warnings 0
AnalysisBase: number of compilation errors 0, warnings 0
AthGeneration: number of compilation errors 0, warnings 0
For experts only: Jenkins output [CI-MERGE-REQUEST-CC7 17362]

Ok looking at the errors perhaps we need to wait for !34872 (merged) to go in so there is not dependency from that side ?

The config is not obvious but I assume if we do not the FieldSvc from there it might be fine

Jenkins please retry a build

This merge request affects 3 packages:

• Trigger/TrigAlgorithms/TrigFastTrackFinder
• Trigger/TrigTools/TrigInDetToolInterfaces
• Trigger/TrigTools/TrigInDetTrackFitter

Adding @sutt ,@nagano ,@bernius as watchers

CI Result FAILURE (hash de8fc090)

	Athena	AthSimulation	AnalysisBase	AthGeneration
externals
cmake
make
required tests
optional tests

Full details available on this CI monitor view
Athena: number of compilation errors 0, warnings 0
AthSimulation: number of compilation errors 0, warnings 0
AnalysisBase: number of compilation errors 0, warnings 0
AthGeneration: number of compilation errors 0, warnings 0
For experts only: Jenkins output [CI-MERGE-REQUEST-CC7 17379]

The newest logs have a hard segmentation relating to getUnbiasedResiduals

stack trace:
RDOtoRDOTrigger 00:19:19  0x7fb5990e4af1 Athena::DebugAids::stacktrace(int) ??:0   + 0x51 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libCxxUtils.so D[0x2caf1]]
RDOtoRDOTrigger 00:19:19  0x7fb5990e65cd Athena::Signal::fatalDump(int, siginfo_t*, void*) ??:0   + 0x21d [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libCxxUtils.so D[0x2e5cd]]
RDOtoRDOTrigger 00:19:19  0x7fb5990e68bd Athena::Signal::fatal(int, siginfo_t*, void*) ??:0   + 0x19d [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libCxxUtils.so D[0x2e8bd]]
RDOtoRDOTrigger 00:19:19  0x7fb597b49bb0 CoreDumpSvcHandler::action(int, siginfo_t*, void*) ??:0   + 0x3e0 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libAthenaServices.so D[0x2d6bb0]]
RDOtoRDOTrigger 00:19:19  0x7fb5b8d2d630 __restore_rt ??:0   + 0x7fb5b8d2d630 [/lib64/libpthread.so.0 D[0xf630]]
RDOtoRDOTrigger 00:19:19  0x7fb592c284f0 IdDictFieldImplementation::unpack(Identifier) const ??:0   + 0x0 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTileConditionsLib.so D[0x1fd4f0]]
RDOtoRDOTrigger 00:19:20  0x7fb56afeda29 TrigInDetTrackFitter::getUnbiasedResiduals(Trk::Track const&, std::vector<TrigL2HitResidual, std::allocator<TrigL2HitResidual> >&, EventContext const&) const ??:0   + 0x869 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTrigInDetTrackFitter.so D[0x69a29]]
RDOtoRDOTrigger 00:19:20  0x7fb573bd3a36 TrigFastTrackFinder::runResidualMonitoring(Trk::Track const&, EventContext const&) const ??:0   + 0x1996 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTrigFastTrackFinder.so D[0xf3a36]]
RDOtoRDOTrigger 00:19:20  0x7fb573bdbdbc TrigFastTrackFinder::fillMon(DataVector<Trk::Track, DataModel_detail::NoBase> const&, TrigVertexCollection const&, TrigRoiDescriptor const&, EventContext const&) const ??:0   + 0x1d6c [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTrigFastTrackFinder.so D[0xfbdbc]]
RDOtoRDOTrigger 00:19:20  0x7fb573bde64d TrigFastTrackFinder::findTracks(InDet::SiTrackMakerEventData_xk&, TrigRoiDescriptor const&, DataVector<Trk::Track, DataModel_detail::NoBase>&, EventContext const&) const ??:0   + 0x1cad [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTrigFastTrackFinder.so D[0xfe64d]]
RDOtoRDOTrigger 00:19:20  0x7fb573bdf763 TrigFastTrackFinder::hltExecute(HLT::TriggerElement const*, HLT::TriggerElement*) ??:0   + 0x123 [/build2/ci-builds/master/Athena/build/Athena/x86_64-centos7-gcc8-opt/lib/libTrigFastTrackFinder.so D[0xff763]]

added review-user-action-required label

added 2 commits

• f1160686 - Merge branch 'mucomb_migrate_magfield' into 'master'
• 80b8ff48 - Move the initialisation to before the return statement

Compare with previous version

Figured it out :

StatusCode TrigInDetTrackFitter::initialize()
{
  ATH_CHECK(m_trackMaker.retrieve());
  if (m_correctClusterPos) {
    ATH_CHECK(m_ROTcreator.retrieve());
  }
  ATH_CHECK( m_fieldCondObjInputKey.initialize());
  return StatusCode::SUCCESS;

  ATH_CHECK(detStore()->retrieve(m_idHelper, "AtlasID"));
  ATH_CHECK(detStore()->retrieve(m_pixelId, "PixelID")); 
  ATH_CHECK(detStore()->retrieve(m_sctId, "SCT_ID"));  
}

-> this, surprisingly, compiles without any warnings, despite the fact that the retrieve statements are after the return. Fixed in latest commit.

removed review-user-action-required label

This merge request affects 4 packages:

• Trigger/TrigAlgorithms/TrigFastTrackFinder
• Trigger/TrigAlgorithms/TrigmuComb
• Trigger/TrigTools/TrigInDetToolInterfaces
• Trigger/TrigTools/TrigInDetTrackFitter

Adding @sutt ,@nagano ,@bernius as watchers

CI Result FAILURE (hash 80b8ff48)

	Athena	AthSimulation	AnalysisBase	AthGeneration
externals
cmake
make
required tests
optional tests

Full details available on this CI monitor view
Athena: number of compilation errors 0, warnings 0
AthSimulation: number of compilation errors 0, warnings 0
AnalysisBase: number of compilation errors 0, warnings 0
AthGeneration: number of compilation errors 0, warnings 0
For experts only: Jenkins output [CI-MERGE-REQUEST-CC7 17459]

Only has known test failures - could do with L1/L2 review?

Hi for the conditions part seems a bit textbook based on all other migrations . One bordeline comment is

void fit(const TrackCollection&, TrackCollection&, const EventContext&, const Trk::ParticleHypothesis& matEffects = Trk::pion) const;

usually you would put const EventContext& 1st https://acode-browser1.usatlas.bnl.gov/lxr/source/athena/Tracking/TrkFitter/TrkFitterInterfaces/TrkFitterInterfaces/ITrackFitter.h or last as you do . Do not mind middle too much though ....

since this is legacy do not know if you want to bother too much.

So assuming that is anyhow legacy and since most of the conditions changes look similar to all the other cases anyhow do not see any issues.

It is the last in a long chain .

For the above comment leave you up to you as positioning of the EventContext argument can turn to bike shed discussion and not sure it even matters here.

added review-approved label

removed review-pending-level-1 label

This isn't legacy, we use TrigInDetTrackFitter for the first-pass fit in FastTrackFinder, the version we plan to run online in Run 3.

So I can move this, it was just because we have the default argument on ParticleHypothesis

If you think it matters put it 1st like the ITrackFitter, that has the same issue for the last ... i.e ParticleHypothesis

But I will not insist too much ...

OK, I'm not convinced it matters :-)

Ok, the rest seem similar to what was done in the Propagators , extrapolator and Fitters elsewhere. There is this state that now also has the mag field cache, and then you work with it in a way that seems pretty reasonable to me .

merged

mentioned in commit 7145fdc8

added sweep:ignore label