feat: Write track parameter performance from track states in a geometry region - #5882
Open
andiwand wants to merge 8 commits into
Open
feat: Write track parameter performance from track states in a geometry region#5882andiwand wants to merge 8 commits into
andiwand wants to merge 8 commits into
Conversation
github-actions
Bot
added
Infrastructure
andiwand
marked this pull request as draft
August 14, 2026 11:12
Moves the track parameters of a track container onto a common surface, typically a perigee, on `Acts::extrapolateTrackToReferenceSurface`. That is where a seed estimate has to be compared to truth: the estimate sits on the bottom space point's sensor, and expressing the truth particle there means intersecting it on a straight line, which ignores the bending in between. The track states are carried over, so they keep the parameters on their own surfaces and the output has the same layering a fitter produces. Tracks whose extrapolation fails are dropped, hence the truth matching has to run downstream of this. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FxPc2jP43AQuQKz8c4qsaC
`addSeeding(trackParameterPerformance=True)` extrapolates the seed tracks to the perigee and writes residuals and pulls of the seed estimate against truth, through `RootTrackParameterPerformanceWriter`. Seed tracks only carry parameters on their innermost state, so the extrapolation uses `TrackExtrapolationStrategy::first`; `firstOrLast` would touch the outermost state, which has none. Nothing is enabled by default. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FxPc2jP43AQuQKz8c4qsaC
`addTrackParameterPerformanceWriter` takes a `resPlotToolConfig`. The default residual axes are cut for fitted tracks, and a seed estimate needs wider ones. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PdoihK1K7tHeLCstKM5ZaB
The `fill` overload taking a truth particle intersects it onto the fitted reference surface, which only makes sense when that surface is a perigee. A second overload takes the truth as `Acts::BoundTrackParameters` already expressed on the same surface, so a per-sensor comparison can use the same histograms. Both funnel into a common private `fill`, and the binning quantities stay derived from each overload's own truth source rather than round-tripping the direction through phi/theta. Missing binning keys and a `paramNames` of the wrong length now throw with a message naming the key, since renaming the local parameters means renaming their residual axes too. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012qsBZHi1zxiUBY5txvM9fg
`truthParametersOnSurface` builds the truth bound parameters on a surface from the simulated hits of a measurement, the same averaging `RootTrackParameterWriter` does inline. `recoParametersOnSurface` picks the parameters of a track state, either a requested `TrackParameterType` or the best available ones. Both are needed to compare a track state to truth on its own surface. Nothing calls them yet. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012qsBZHi1zxiUBY5txvM9fg
`RootTrackParameterPerformanceWriter` gains `parameterSource = TrackState`, which compares every measurement state to the truth on its own surface instead of comparing the track reference parameters to the truth particle. `geometrySelection` restricts that to a geometry region, and `parameterType` picks predicted, filtered, smoothed, or unbiased parameters. `addTrackStateParameterPerformanceWriter` wires it up from python. Nothing is enabled by default. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012qsBZHi1zxiUBY5txvM9fg
andiwand
force-pushed
the
track-state-parameter-performance
branch
from
August 14, 2026 11:23
8a96ac2 to
af65c24
Compare
This was referenced Aug 14, 2026
github-actions
Bot
removed
the
Infrastructure
Contributor
Public API surface diffNo change to the public API surface. ✅ |
Contributor
yannikaMatt
pushed a commit
to yannikaMatt/acts
that referenced
this pull request
Aug 21, 2026
…5721) `addSeeding(trackParameterPerformance=True)` extrapolates the seed tracks to a perigee and writes residuals and pulls of the seed estimate against truth, through `RootTrackParameterPerformanceWriter`. Seed parameters steer the CKF, and nothing so far plots them next to the fitter performance output. The perigee is what makes the comparison meaningful, since `ResPlotTool` intersects the truth particle onto the reference surface along a straight line and a 1 GeV track in 2 T has already bent by ~18 mrad at the bottom space point. Seed tracks carry parameters only on their innermost state, so the extrapolation uses `TrackExtrapolationStrategy::first`. Nothing is enabled by default. The per-state half follows in acts-project#5882. --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
yannikaMatt
pushed a commit
to yannikaMatt/acts
that referenced
this pull request
Aug 21, 2026
…5721) `addSeeding(trackParameterPerformance=True)` extrapolates the seed tracks to a perigee and writes residuals and pulls of the seed estimate against truth, through `RootTrackParameterPerformanceWriter`. Seed parameters steer the CKF, and nothing so far plots them next to the fitter performance output. The perigee is what makes the comparison meaningful, since `ResPlotTool` intersects the truth particle onto the reference surface along a straight line and a 1 GeV track in 2 T has already bent by ~18 mrad at the bottom space point. Seed tracks carry parameters only on their innermost state, so the extrapolation uses `TrackExtrapolationStrategy::first`. Nothing is enabled by default. The per-state half follows in acts-project#5882. --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
…er-performance # Conflicts: # Examples/Algorithms/Utilities/include/ActsExamples/Utilities/TrackExtrapolationAlgorithm.hpp # Examples/Algorithms/Utilities/src/TrackExtrapolationAlgorithm.cpp # Python/Examples/python/reconstruction.py # Python/Examples/src/Utilities.cpp
Contributor
Author
|
Notes for review, moved out of the description now that it is the commit The per-state half of the original #5721, which was rewritten down to the Notes
ValidationMeasured before the merge of Generic detector, 20 events, truth-smeared seeding into the KF.
|
|
Define `truthParametersOnSurface` and `recoParametersOnSurface` with their `ActsExamples::` qualification instead of re-opening the namespace, so a signature that drifts from the header fails to compile rather than silently becoming a new overload. Also drop the unused `addTrackStateParameterPerformanceWriter` and `defaultOnSurfaceResPlotToolConfig` helpers. The writer stays reachable from Python through its bindings. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013asNjmmrVRex58ppaX1WJ4
andiwand
marked this pull request as ready for review
August 24, 2026 16:01
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
RootTrackParameterPerformanceWritergainsparameterSource = TrackState: itcompares every measurement state of a track to the truth on the surface that
state sits on, instead of the track reference parameters to the truth particle.
geometrySelectionnarrows that to a geometry region, so a single volume orlayer can be looked at on its own, and
parameterTypepicks predicted,filtered, smoothed or unbiased parameters.
Comparing on the state's own surface is what makes a per-sensor estimate
measurable at all. Extrapolating to a common reference first, as
TrackExtrapolationAlgorithmdoes, answers a different question; this is thelocal half.
Supporting it:
truthParametersOnSurfaceandrecoParametersOnSurfacein thenew
ParametersOnSurface.hpp, aResPlotTool::filloverload taking the truthas bound parameters on the fitted surface, and
addTrackStateParameterPerformanceWriterin python, whosedefaultOnSurfaceResPlotToolConfignames the local parametersloc0/loc1rather than
d0/z0, which only mean something on a line surface.ResPlotToolnow names the offending key whenvarBinningorparamNamesdoes not cover the bound parameters.
Nothing is enabled by default.