Add dt_hit / dt_ptoca linear fit information for hits on a track

[michaelmackenzie]

The straw hit time vs. track expected time at the hit position has a non-unit slope when the wrong mass hypothesis is assumed. This slope information can be used in PID, as is relevant to PR Mu2e/MLTrain#9. This adds a reco data product with this information and a producer module to create a collection of these products for input KalSeed or KalSeedPtr collections.

[FNALbuild]

Hi @michaelmackenzie,
You have proposed changes to files in these packages:

• ParticleID
• RecoDataProducts

which require these tests: build.

@Mu2e/write, @Mu2e/fnalbuild-users have access to CI actions on main.

⌛ The following tests have been triggered for e7ca770: build (Build queue - API unavailable)

About FNALbuild. Code review on Mu2e/Offline.

[FNALbuild]

☔ The build tests failed for e7ca770.

Test	Result	Details
test with	✅	Command did not list any other PRs to include
merge	✅	Merged e7ca770 at 3d84f93
build (prof)	✅	Log file. Build time: 04 min 18 sec
ceSimReco	✅	Log file.
g4test_03MT	✅	Log file.
transportOnly	✅	Log file.
POT	✅	Log file.
g4study	✅	Log file.
cosmicSimReco	✅	Log file.
cosmicOffSpill	✅	Log file.
ceSteps	✅	Log file.
ceDigi	✅	Log file.
muDauSteps	✅	Log file.
ceMix	✅	Log file.
rootOverlaps	✅	Log file.
g4surfaceCheck	✅	Log file.
trigger	✅	Log file.
check_cmake	❌	Log file. Return Code 1.
FIXME, TODO	✅	TODO (0) FIXME (0) in 2 files
clang-tidy	➡️	2 errors 9 warnings
whitespace check	✅	no whitespace errors found

N.B. These results were obtained from a build of this Pull Request at e7ca770 after being merged into the base branch at 3d84f93.

For more information, please check the job page here.
Build artifacts are deleted after 5 days. If this is not desired, select Keep this build forever on the job page.

[oksuzian]

Possible Improvements

The following are concrete, diff-anchored points worth addressing before merge:

1. Weight formula looks physically wrong / risks division-by-zero.
   ParticleID/src/TrackDtDt_module.cc:60-62

   const double t_unc  = hit._udresid / hit._dvel;
   const double weight = 1./t_unc;

   _udresid is the unbiased distance residual value (see TrkStrawHitSeed.hh:186), not an uncertainty — so residual/velocity is not a time uncertainty. A time error would more plausibly come from a variance term (e.g. sqrt(_udresidmvar)/_dvel, or _utocavar). As written, weight can be negative (residuals are signed) and 1/t_unc blows up when _udresid == 0 or _dvel == 0 (the legacy BTrk constructor sets _dvel = 0). Please confirm the intended uncertainty and guard against zero/negative weights.

2. LsqSums2 accessors are non-const, but you pass seed by const-ref. This compiles only because fitter is a local non-const. Still, consider that dof_ = fitter.qn() - 2 can go negative for tracks with <2 hits, and chisq_ = chi2Dof()*dof_ will then be meaningless. Add a guard for qn() < 3 (skip or set sentinel values).

3. Struct members are uninitialized when set from the fit. KalSeedDtDt's default ctor zero-inits, but fitTrackDtDt creates a result and sets 5 fields — fine today, but if any LsqSums2 accessor returns NaN (e.g. det()==0 for degenerate input), it propagates silently. A quick isnan check before emplace would harden the product.

4. Minor: #include <iostream> guarded diagnostics use std::cout; consider mf::LogInfo/MessageLogger for consistency with framework conventions, though many Mu2e modules do use std::cout under diagLevel.

[oksuzian]

@FNALbuild run build test

[FNALbuild]

⌛ The following tests have been triggered for 036e83b: build (Build queue - API unavailable)

[FNALbuild]

☀️ The build tests passed at 036e83b.

Test	Result	Details
test with	✅	Command did not list any other PRs to include
merge	✅	Merged 036e83b at eb6c796
build (prof)	✅	Log file. Build time: 04 min 20 sec
ceSimReco	✅	Log file.
g4test_03MT	✅	Log file.
transportOnly	✅	Log file.
POT	✅	Log file.
g4study	✅	Log file.
cosmicSimReco	✅	Log file.
cosmicOffSpill	✅	Log file.
ceSteps	✅	Log file.
ceDigi	✅	Log file.
muDauSteps	✅	Log file.
ceMix	✅	Log file.
rootOverlaps	✅	Log file.
g4surfaceCheck	✅	Log file.
trigger	✅	Log file.
check_cmake	✅	Log file.
FIXME, TODO	✅	TODO (0) FIXME (0) in 2 files
clang-tidy	➡️	2 errors 9 warnings
whitespace check	✅	no whitespace errors found

N.B. These results were obtained from a build of this Pull Request at 036e83b after being merged into the base branch at eb6c796.

For more information, please check the job page here.
Build artifacts are deleted after 5 days. If this is not desired, select Keep this build forever on the job page.

[michaelmackenzie]

I updated the dt/dt slope definition to match Dave's suggestion and added debug histogramming for easy testing. I tested on flat electron and muon inputs with an electron track fit that the slope distribution is peaked at 1 for true electrons and ~1.5 for true muons.

[brownd1978]

This looks good Michael.
What is your plan for adding this to the reconstruction sequence?

[michaelmackenzie]

This looks good Michael. What is your plan for adding this to the reconstruction sequence?

I'm not positive about the best location. It could either go in reco after track reconstruction or in the ntupling stage before the MVA consumers. I don't have a strong opinion on this, but we can put this in reco if we want to minimize any reconstruction-type producers within ntupling jobs. Do you have a preference or alternate suggestion?

[brownd1978]

Putting it in the general track reco sequence will add it to every track, I'm not sure if we need that. Putting it in the PID sequence seems more targeted.