Navigation performance benchmark - implementing #49 #530

MoritzNeuberger · 2025-11-10T10:01:29Z

This is an implementation of CJs benchmarking class referenced in #49.
Additionally, I added an optional output to report the median time per grid point.

Progress:

benchmark wrapper for remage as an additional cli
- bare-bones cli implemented: given geometry samples from world volume, generates simple statistics and plots
- gdml generator: given complex gdml, extract parts of interest and generate world volume with sufficient buffer
- better thought into which plots/statistics: what is relevant for the geometry generation process
overlap identifier, i.e., check how many surfaces the Geantino passes. If not even, there must be an overlap [link] <-- necessary?

Copilot

Pull Request Overview

This PR adds a new benchmark generator (RMGGeneratorBenchmark) for measuring geometry navigation performance in Geant4 simulations. The generator systematically samples particles across three orthogonal planes (XZ, YZ, XY) using a grid pattern and records timing data to identify potential geometry bottlenecks.

Key Changes:

Implements RMGGeneratorBenchmark class that shoots geantinos through the detector geometry in a systematic grid pattern
Adds RMGBenchmarkOutputScheme to store benchmark timing data in three separate ntuples (one per plane)
Integrates the benchmark generator into the existing generator framework with CLI configuration options

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 20 comments.

Show a summary per file

File	Description
`src/RMGMasterGenerator.cc`	Adds benchmark generator to the generator factory switch statement
`src/RMGGeneratorBenchmark.cc`	Implements the benchmark generator with grid-based particle sampling and timing measurement
`src/RMGBenchmarkOutputScheme.cc`	Implements output scheme for storing benchmark data to three auxiliary ntuples
`src/CMakeLists.txt`	Registers new source and header files in build system
`include/RMGUserInit.hh`	Registers benchmark output scheme as an optional output scheme
`include/RMGMasterGenerator.hh`	Adds `kBenchmark` enum value to Generator enumeration
`include/RMGGeneratorBenchmark.hh`	Header file declaring the benchmark generator class and its members
`include/RMGBenchmarkOutputScheme.hh`	Header file declaring the benchmark output scheme class

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

include/RMGGeneratorBenchmark.hh

src/RMGGeneratorBenchmark.cc

include/RMGGeneratorBenchmark.hh

src/RMGGeomBench.cc

src/RMGGeneratorBenchmark.cc

gipert

Thanks Mo, I had a quick look and left some first comments.

I think just naming things with "benchmark" is not enough, we should also say "geometry". I would use everywhere either GeometryBenchmark or GeoBenchmark (if you prefer shorter names)

Also, the 3D thing is quite some code and I'm not convinced that it's useful... maybe this can come in a future PR

include/RMGBenchmarkOutputScheme.hh

include/RMGGeneratorBenchmark.hh

python/remage/geombench/__init__.py

python/remage/geometry_benchmark/__main__.py

python/remage/geometry_benchmark/cli.py

ManuelHu · 2025-11-13T09:52:21Z

I think just naming things with "benchmark" is not enough, we should also say "geometry". I would use everywhere either GeometryBenchmark or GeoBenchmark (if you prefer shorter names)

I was thinking about names. We could maybe use an even shorter name like GeomBench, but use that everywhere (in the C++ file and class names, geombench as python subpackage, remage-geombench as the executable name, ...).

python/remage/geometry_benchmark/cli.py

python/remage/geombench/__init__.py

python/remage/geometry_benchmark/__main__.py

gipert · 2025-11-13T09:53:56Z

I was thinking about names. We could maybe use an even shorter name like GeomBench, but use that everywhere (in the C++ file and class names, geombench as python subpackage, remage-geombench as the executable name, ...).

yes I agree. I care more about consistency than the exact name.

MoritzNeuberger · 2025-12-02T09:03:32Z

Implemented the suggestions consisting of:

change of names: now the RMG classes and python package are all consistently named geombench
simplified RMG calculations using G4ThreeVectors, also using size_t more
remove feature from python package not needed for barebones release
added interface for cli for batch use

In addition to that, there was also an internal calculation change:

Instead of calculating all batches for a pixel one after the other, the orientation was flipped, meaning the first batch for each pixel is calculated one after the other before switching to the next batch. This significantly reduces spikes in the output.

I will now focus on implementing the gdml generator as outlined above.
(A preliminary draft of my thesis is in Stefan's inbox, so I have some time at the moment.)

gipert · 2025-12-02T10:40:44Z

I had to approve the workflow, apparently giving you "triage" permissions does not mean you are "trusted". We should look into this, it's a bit annoying.

ManuelHu · 2025-12-02T10:55:33Z

No, I think this is because the last commit was by pre-commit and that is not a user with a repository role?

gipert · 2025-12-02T10:57:48Z

Ah ok, good then. Maybe we can also give it triage permissions? Not sure it's possible

MoritzNeuberger · 2025-12-02T12:21:06Z

Added a gdml_handling.py file. It contains functionality to extract individual lv of a complex gdml (including definitions and material information) as well as functionality to place an object of interest in a world volume with adjustable buffer space. In the default situation, the CLI just takes the provided GDML file and adds a 12.5% buffer on each side (size adjustable in CLI with buffer-fraction). With the additional CLI argument object-lv-name, one can specify a specific logical volume that is then extracted, including its daughters. This is then placed again in a world volume with a buffer and analyzed.

python/remage/geombench/gdml_handling.py

python/remage/geombench/cli.py

MoritzNeuberger · 2025-12-02T13:18:03Z

I will add some documentation while waiting for further suggestions.

MoritzNeuberger · 2025-12-02T13:46:24Z

I guess this PR should not yet include an overlap identifier. Though since RMGGeometryCheckOutputScheme has very similar functionality, I do think we should definitely combine them at a later point.

MoritzNeuberger · 2025-12-02T14:54:12Z

Added terminal histogram output with histprint from scikit (backend for the terminal printing in hist).

Also, changed so that if there is already a box world volume, that one is enlarged instead of nesting previous world volumes inside one another.

gipert · 2025-12-02T14:55:53Z

Could you rewrite the git history, tidy up and remove pre-commit bot commits? If you let the bot push them, the CI needs approval to run...

MoritzNeuberger · 2025-12-02T14:57:22Z

With the changes in the ordering of which batch/pixel simulation order, the output is not very robust to spikes (no more artifacts in the visualization!)

The histogram is still quite ugly at the moment. It's supposed to be log scale, but I guess that is a bit awkward with this package ... can be fine-tuned though.

                                                                                       Median Simulation Times per Event [μs]                                                                                       
  0.030 _ x 10^+02                                                                                                                                                                                       60458/row ╷
  0.031 _███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████
  0.032 _
  0.034 _
  0.035 _████████████████████████████████████████
  0.037 _
  0.038 _
  0.040 _
  0.041 _█████████████████████████████████████████████████████████
  0.043 _
  0.045 _
  0.046 _
  0.048 _
  0.050 _██████
  0.052 _
  0.054 _
  0.057 _
  0.059 _
  0.061 _██
  0.064 _
  0.066 _
  0.069 _
  0.072 _
  0.075 _
  0.078 _
  0.081 _█████████████████████████████████
  0.084 _
  0.088 _██████████
  0.091 _██████████████████████████
  0.095 _
  0.099 _██████
  0.103 _████████████████████████████████
  0.107 _█████
  0.111 _█████████████████████████████████████
  0.115 _████
  0.120 _███████████
  0.125 _██
  0.130 _████
  0.135 _███
  0.141 _██████████
  0.147 _████
  0.152 _████████████
  0.159 _███
  0.165 _██████████████
  0.172 _██████
  0.179 _██
  0.186 _██████████
  0.193 _████
  0.201 _████████████
  0.209 _███
  0.218 _████████████████
  0.227 _██████████
  0.236 _███
  0.245 _████
  0.255 _███
  0.266 _█
  0.276 _█
  0.287 _
  0.299 _
  0.311 _█
  0.324 _
  0.337 _

gipert · 2025-12-02T15:02:37Z

very ugly... if there is no horizontal option (like in Julia's BenchmarkTools) i would skip this

MoritzNeuberger · 2025-12-02T15:04:03Z

Not with this package AFAIK histoprint

MoritzNeuberger · 2025-12-02T15:07:56Z

This package can generate sparklines: deeplook/sparklines

gipert · 2025-12-02T15:10:02Z

watch out when you add dependencies: they need to be on conda-forge and pypi

gipert · 2025-12-02T15:11:46Z

anyways now that i think about it i would not add another dependency just for a fancy plot...

MoritzNeuberger · 2025-12-02T15:21:19Z

FYI this is what would be possible with the above package:

================================================================================
Median Simulation Times per Event [μs]
================================================================================
   █                                              
   █                                              
   █                                              
▃  █                                              
█▁ █                                              
██ █         ▆                                    
██ █         ██ █▆      ▁▁                        
██▁█▂▁▃▁▃▁▁▁▁██▃██▇▇▇█████▄▃▂▂▁▁▁▁▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁

|           |            |           |           |
3.00        7.94         22.78       65.38       173.00

Summary Statistics:
  Count: 187500
  Min:   3.00 μs
  Max:   173.00 μs
  Mean:  9.62 μs
  Std:   8.44 μs
================================================================================

If you are not interested, no worries.

style: pre-commit fixes style: pre-commit fixes Update include/RMGGeneratorBenchmark.hh Co-authored-by: Copilot <[email protected]> Update include/RMGGeneratorBenchmark.hh Co-authored-by: Copilot <[email protected]> Implementing suggestions

…n calculation time Update python/remage/geometry_benchmark/cli.py Co-authored-by: Manuel Huber <[email protected]>

Introduces logic to extract and benchmark individual logical volumes from complex GDML geometries, including new CLI options for selecting a component and specifying buffer space around the geometry. Refactors geometry handling to wrap extracted components in a temporary, buffered world volume for isolated benchmarking. Improves summary generation to reflect component-specific output naming and corrects extent calculation for visualization. Enhances sampling region logging for clarity on simulation setup. style: pre-commit fixes Incorporating Luigis suggestions style: pre-commit fixes

The text is ai generated, but I checked it and through out stuff when it was yapping. style: pre-commit fixes fix docstring? style: pre-commit fixes Improves output naming and adds simulation time histograms Standardizes output file naming by introducing an explicit file stem variable, ensuring more predictable and descriptive output filenames, especially when extracting logical volumes. Refactors related CLI and summary generation logic to use this file stem throughout. Expands world volume resizing to avoid unnecessary registry duplication and ensure correct world volume updates. Adds a simulation time histogram printout using histoprint for better performance diagnostics. Includes minor doc and formatting fixes for clarity. style: pre-commit fixes style: pre-commit fixes precommit skip

…tions fixed some stuff

…ritzNeuberger/remage into navigaion_benchmark_utilities

python/CMakeLists.txt

python/remage/geombench/cli.py

python/remage/geombench/gdml_handling.py

python/remage/geombench/summary_generator.py

src/RMGGeomBench.cc

gipert · 2025-12-05T17:16:03Z

I had a quick look, i think there is a lot of unnecessary logic (this is typical of the AI). could you review and try to simplify? even if it works, we should try to keep this code maintainable

- Refactor geombench CLI and GDML handling, removing unnecessary complexity in util functions, removed macro file generation, piping it directly to the cli - Changed Out to OutFormat where applicable - Left try-except logic in to delte the tmp gdml file if generated, can be removed in the future when option to pipe gdml to the cli is possible

… plotting

MoritzNeuberger · 2025-12-09T14:33:49Z

I adjusted the plotting to use hist. Specifically, it generates a hist.Hist with two axes and then manually enters the values into the content. The plotting is done via plot2d. This requires hist and (AFAIK) mplhep to be added as dependencies.

I gave up on using the hist generated color bar since I could not get the z-label implemented.

The rendering looks like this now:

MoritzNeuberger · 2025-12-18T18:12:12Z

Are there any points not considered yet?

ManuelHu

just some minor comments remaining for me

python/remage/geombench/cli.py

src/RMGGeomBench.cc

.pre-commit-config.yaml

python/remage/geombench/gdml_handling.py

include/RMGGeomBench.hh

src/RMGGeomBenchOutputScheme.cc

ManuelHu · 2025-12-19T10:04:44Z

src/remage-doc-dump.cc

  new RMGIsotopeFilterScheme();
  new RMGTrackOutputScheme();
  new RMGParticleFilterScheme();
+  new RMGGeomBenchOutputScheme();


this doesn't even define commands, so you can remove this here and also the DefineCommands in RMGGeomBenchOutputScheme...?

ManuelHu

additional comments from clang-tidy:

/work/remage/src/RMGGeomBench.cc:26:1: warning: constructor does not initialize these fields: totalnevents, totalnpixels, npixelsperrow, neventsperpixel, cubesize, npixels_x, npixels_y, npixels_z, ID, starttime, currenttime, bunchstarttime, events_per_bunch, total_batch_rounds, current_batch_event, current_pixel_index, current_batch_round [cppcoreguidelines-pro-type-member-init]
   26 | RMGGeomBench::RMGGeomBench() : RMGVGenerator("Benchmark") {
      | ^
/work/remage/src/RMGGeomBench.cc:115:15: warning: redundant string initialization [readability-redundant-string-init]
  115 |   std::string targetvolumename = "";
      |               ^~~~~~~~~~~~~~~~~~~~~
      |               targetvolumename
/work/remage/src/RMGGeomBench.cc:226:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  226 |   int pixels_xz = npixels_x * npixels_z;
      |                   ^
/work/remage/src/RMGGeomBench.cc:227:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  227 |   int pixels_yz = npixels_y * npixels_z;
      |                   ^
/work/remage/src/RMGGeomBench.cc:228:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  228 |   int pixels_xy = npixels_x * npixels_y;
      |                   ^
/work/remage/src/RMGGeomBench.cc:235:21: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  235 |   neventsperpixel = totalnevents / totalnpixels;
      |                     ^
/work/remage/src/RMGGeomBench.cc:329:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  329 |         max_i = npixels_x;
      |                 ^
/work/remage/src/RMGGeomBench.cc:330:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  330 |         max_j = npixels_z;
      |                 ^
/work/remage/src/RMGGeomBench.cc:333:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  333 |         max_i = npixels_y;
      |                 ^
/work/remage/src/RMGGeomBench.cc:334:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  334 |         max_j = npixels_z;
      |                 ^
/work/remage/src/RMGGeomBench.cc:337:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  337 |         max_i = npixels_x;
      |                 ^
/work/remage/src/RMGGeomBench.cc:338:17: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  338 |         max_j = npixels_y;
      |                 ^
/work/remage/src/RMGGeomBench.cc:345:16: warning: variable 'x_pos' is not initialized [cppcoreguidelines-init-variables]
    4 |         double x_pos, y_pos, z_pos;
      |                ^    
      |                      = NAN
/work/remage/src/RMGGeomBench.cc:345:23: warning: variable 'y_pos' is not initialized [cppcoreguidelines-init-variables]
  345 |         double x_pos, y_pos, z_pos;
      |                       ^    
      |                             = NAN
/work/remage/src/RMGGeomBench.cc:345:30: warning: variable 'z_pos' is not initialized [cppcoreguidelines-init-variables]
  345 |         double x_pos, y_pos, z_pos;
      |                              ^    
      |                                    = NAN
/work/remage/src/RMGGeomBench.cc:374:18: warning: variable 'median_time' is not initialized [cppcoreguidelines-init-variables]
  374 |           double median_time;
      |                  ^          
      |                              = NAN
/work/remage/src/RMGGeomBench.cc:427:23: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  427 |   int events_so_far = ID;
      |                       ^
/work/remage/src/RMGGeomBench.cc:428:25: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  428 |   current_batch_round = events_so_far / (totalnpixels * events_per_bunch);
      |                         ^
/work/remage/src/RMGGeomBench.cc:429:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  429 |   int remainder = events_so_far % (totalnpixels * events_per_bunch);
      |                   ^
/work/remage/src/RMGGeomBench.cc:438:30: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  438 |       int prev_pixel_index = (ID - 1) % (totalnpixels * events_per_bunch) / events_per_bunch;
      |                              ^
/work/remage/src/RMGGeomBench.cc:458:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  458 |   int pixels_xz = npixels_x * npixels_z;
      |                   ^
/work/remage/src/RMGGeomBench.cc:459:19: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  459 |   int pixels_yz = npixels_y * npixels_z;
      |                   ^
/work/remage/src/RMGGeomBench.cc:461:7: warning: variable 'plane' is not initialized [cppcoreguidelines-init-variables]
  461 |   int plane;
      |       ^    
      |             = 0
/work/remage/src/RMGGeomBench.cc:462:7: warning: variable 'i_index' is not initialized [cppcoreguidelines-init-variables]
  462 |   int i_index, j_index;
      |       ^      
      |               = 0
/work/remage/src/RMGGeomBench.cc:462:16: warning: variable 'j_index' is not initialized [cppcoreguidelines-init-variables]
  462 |   int i_index, j_index;
      |                ^      
      |                        = 0
/work/remage/src/RMGGeomBench.cc:467:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  467 |     i_index = current_pixel_index % npixels_x;
      |               ^
/work/remage/src/RMGGeomBench.cc:468:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  468 |     j_index = current_pixel_index / npixels_x;
      |               ^
/work/remage/src/RMGGeomBench.cc:473:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  473 |     i_index = local_idx % npixels_y;
      |               ^
/work/remage/src/RMGGeomBench.cc:474:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  474 |     j_index = local_idx / npixels_y;
      |               ^
/work/remage/src/RMGGeomBench.cc:479:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  479 |     i_index = local_idx % npixels_x;
      |               ^
/work/remage/src/RMGGeomBench.cc:480:15: warning: narrowing conversion from 'size_t' (aka 'unsigned long') to signed type 'int' is implementation-defined [cppcoreguidelines-narrowing-conversions]
  480 |     j_index = local_idx / npixels_x;
      |               ^

Co-authored-by: Manuel Huber <[email protected]>

…gaion_benchmark_utilities

…ritzNeuberger/remage into navigaion_benchmark_utilities

Co-authored-by: Manuel Huber <[email protected]>

…ritzNeuberger/remage into navigaion_benchmark_utilities

… "X", "Y", "Z", and "Time" to "x", "y", "z", and "time" respectively.

gipert

I also have few minor comments. Thanks a lot for the work Moritz!

gipert · 2025-12-19T14:38:40Z

docs/manual/geombench.md

+### Key metrics
+
+**Mean navigation time** : Average time spent navigating to each point. Lower is
+better.
+
+**Max navigation time** : Slowest navigation time encountered. High values
+indicate geometry hotspots.
+
+**Standard deviation** : Variability in navigation time. High values suggest
+non-uniform complexity.


I think this can be removed.

gipert · 2025-12-19T14:40:28Z

docs/manual/geombench.md

+3. **Material boundaries**: Excessive material changes force more boundary
+   checks. Consolidate materials where physics allows.
+
+4. **Envelope optimization**: Proper envelope volumes can accelerate navigation


I think you don't explain well what you mean here

gipert · 2025-12-19T14:42:43Z

python/remage/geombench/cli.py

+        - output_dir: Directory to store benchmark output files.
+        - num_events: Number of events to simulate.
+        - grid_increment: Increment between grid points in mm.
+        - grid_increments: Optional dict of increments per dimension in mm.
+        - dry_run: If set, only generate the macro file without running remage.


this does not render correctly in the API docs

gipert · 2025-12-19T14:44:29Z

python/remage/geombench/cli.py

+            msg = f"{key}: {value}"
+            logger.info(msg)
+
+    except Exception as e:


is this try-catch block really needed?

gipert · 2025-12-19T16:57:00Z

python/remage/geombench/gdml_handling.py

+
+import pyg4ometry
+
+## GDML Handling Functions - Helper utilities for copying registry resources


This should become the module docstring

gipert · 2025-12-19T17:01:55Z

python/remage/geombench/gdml_handling.py

+        Dictionary with keys:
+        - "object_lv": The loaded and renamed logical volume
+        - "registry": The pyg4ometry registry containing the geometry


this also does not render correctly

gipert · 2025-12-19T17:06:03Z

docs/manual/geombench.md

+Geometry Benchmark Analysis Results:
+mean_navigation_time: 2.45 ns
+median_navigation_time: 1.89 ns
+std_navigation_time: 1.23 ns
+max_navigation_time: 45.67 ns
+min_navigation_time: 0.34 ns


is this the actual output?

Copilot AI review requested due to automatic review settings November 10, 2025 10:01

Copilot AI reviewed Nov 10, 2025

View reviewed changes

gipert linked an issue Nov 10, 2025 that may be closed by this pull request

Implement utility to benchmark navigation performance #49

Open

gipert reviewed Nov 13, 2025

View reviewed changes

ManuelHu reviewed Nov 13, 2025

View reviewed changes

gipert reviewed Dec 2, 2025

View reviewed changes

python/remage/geombench/gdml_handling.py Outdated Show resolved Hide resolved

gipert reviewed Dec 2, 2025

View reviewed changes

python/remage/geombench/cli.py Outdated Show resolved Hide resolved

MoritzNeuberger and others added 5 commits December 2, 2025 16:58

initial commit - merging of CJs classes

e008c6d

Adds geometry benchmark CLI

c3a6ebe

style: pre-commit fixes style: pre-commit fixes Update include/RMGGeneratorBenchmark.hh Co-authored-by: Copilot <[email protected]> Update include/RMGGeneratorBenchmark.hh Co-authored-by: Copilot <[email protected]> Implementing suggestions

implemented better approach to be less suseptable for sudden spikes i…

c36fd05

…n calculation time Update python/remage/geometry_benchmark/cli.py Co-authored-by: Manuel Huber <[email protected]>

MoritzNeuberger added 2 commits December 3, 2025 11:35

Added tests for the python package, also removed some depricated func…

5a725f5

…tions fixed some stuff

Merge branch 'navigaion_benchmark_utilities' of https://github.com/Mo…

0098522

…ritzNeuberger/remage into navigaion_benchmark_utilities

ManuelHu reviewed Dec 3, 2025

View reviewed changes

python/CMakeLists.txt Show resolved Hide resolved

Added suggestion

f90ac28