ref: Use track parameter vector directly in seeding

[niermann999]

Use the track parameter vector class directly in seeding instead of the column matrix bound_vector. This has the advantage, that the assertions in the setters of the track_parameter_vector class in detray are getting checked in debug builds.

Also adds the geometry context to the spacepoint formation interface and prevents the transform3 in the spacepoint formation from calculating an additional cross product, by providing all axes to the constructor.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[stephenswat]

Performance summary

Here is a summary of the performance effects of this PR:

Graphical

Tabular

Kernel	1febe21	04cd604	Delta
fit	280.09 ms	325.59 ms	16.2%
propagate_to_next_surface	118.39 ms	180.55 ms	52.5%
find_tracks	26.35 ms	26.93 ms	2.2%
count_triplets	14.17 ms	14.17 ms	0.0%
find_triplets	5.99 ms	5.98 ms	-0.1%
build_tracks	1.07 ms	1.16 ms	8.7%
find_doublets	896.31 μs	900.02 μs	0.4%
Thrust::sort	778.37 μs	812.35 μs	4.4%
prune_tracks	672.38 μs	750.74 μs	11.7%
ccl_kernel	681.14 μs	682.74 μs	0.2%
count_doublets	627.28 μs	626.72 μs	-0.1%
select_seeds	361.11 μs	362.68 μs	0.4%
apply_interaction	150.70 μs	157.58 μs	4.6%
update_triplet_weights	96.24 μs	95.90 μs	-0.4%
fill_sort_keys	68.27 μs	76.15 μs	11.5%
populate_grid	30.46 μs	30.40 μs	-0.2%
count_grid_capacities	29.21 μs	29.18 μs	-0.1%
estimate_track_params	34.13 μs	21.53 μs	-36.9%
form_spacepoints	13.43 μs	13.50 μs	0.5%
reduce_triplet_counts	6.64 μs	6.62 μs	-0.3%
static_kernel	1.76 μs	1.75 μs	-0.4%
make_barcode_sequence	993.12 ns	995.72 ns	0.3%
fill_prefix_sum	165.48 ns	165.51 ns	0.0%
Total	450.50 ms	558.96 ms	24.1%

Important

All metrics in this report are given as reciprocal throughput, not as wallclock runtime.

Warning

At least one kernel incurred a significant performance regression.

Note

This is an automated message produced on the explicit request of a human being.

[stephenswat]

@niermann999 that doesn't look great! 😨

[stephenswat]

Performance summary

Here is a summary of the performance effects of this PR:

Graphical

Tabular

Kernel	1febe21	04cd604	Delta
fit	279.50 ms	326.86 ms	16.9%
propagate_to_next_surface	118.38 ms	180.03 ms	52.1%
find_tracks	26.39 ms	26.89 ms	1.9%
count_triplets	14.17 ms	14.18 ms	0.1%
find_triplets	5.98 ms	6.00 ms	0.4%
build_tracks	1.07 ms	1.14 ms	5.8%
find_doublets	896.81 μs	904.24 μs	0.8%
Thrust::sort	777.61 μs	811.26 μs	4.3%
prune_tracks	674.27 μs	751.08 μs	11.4%
ccl_kernel	680.71 μs	681.12 μs	0.1%
count_doublets	626.15 μs	628.20 μs	0.3%
select_seeds	361.19 μs	354.16 μs	-1.9%
apply_interaction	150.89 μs	157.78 μs	4.6%
update_triplet_weights	96.54 μs	96.01 μs	-0.6%
fill_sort_keys	68.24 μs	76.14 μs	11.6%
populate_grid	30.39 μs	30.45 μs	0.2%
count_grid_capacities	29.20 μs	29.20 μs	0.0%
estimate_track_params	34.07 μs	21.55 μs	-36.8%
form_spacepoints	13.53 μs	13.41 μs	-0.9%
reduce_triplet_counts	6.63 μs	6.61 μs	-0.4%
static_kernel	1.76 μs	1.76 μs	-0.1%
make_barcode_sequence	987.73 ns	993.45 ns	0.6%
fill_prefix_sum	165.51 ns	165.50 ns	-0.0%
Total	449.93 ms	559.65 ms	24.4%

Important

All metrics in this report are given as reciprocal throughput, not as wallclock runtime.

Warning

At least one kernel incurred a significant performance regression.

Note

This is an automated message produced on the explicit request of a human being.

[niermann999]

@niermann999 that doesn't look great! 😨

Indeed! This is curious, it should not change any results. I will have a look at why this influences track finding