update to detray v51

[beomki-yeo]

No description provided.

[krasznaa]

@beomki-yeo, @niermann999, is this the update necessary for the full-chain? 🤔

It seems to be just a p_mag -> p_tot issue. Will you fix it?

[krasznaa]

As long as the CI is happy, this should go in. As it is required for #509. 😉

[krasznaa]

Unfortunately the test failures are not something that I could/would fix. 😦

[beomki-yeo]

Probably the changes in the detray::navigator are conflicting with traccc::kalman_actor. @niermann999 any thought?

[beomki-yeo]

It might be something else because ckf does not use kalman actor as far as I know

[beomki-yeo]

I believe this will fix the errors in the CI, but I cannot guarantee that this will work for ODD.

[beomki-yeo]

@krasznaa

If you think the ODD test is not working in the current setup, I suggest to remove the detray::stepper_default_policy from the rk_stepper_type. This change should be available for CPU fitting and finding algorithm.

-    using rk_stepper_type = detray::rk_stepper<
-        b_field_t::view_t, typename host_detector_type::transform3,
-        detray::constrained_step<>, detray::stepper_default_policy>;

+   using rk_stepper_type = detray::rk_stepper<
+      b_field_t::view_t, typename host_detector_type::transform3,
+     detray::constrained_step<>>;

[krasznaa]

Unfortunately as long as the tests keep failing, the PR definitely doesn't go in. 🤔

If anything, your latest changes only made things worse. 😦 Previously the failing tests complained about efficiencies going down a little bit. Now the tests don't find any tracks anymore. 😦

At the same time this gives me hope that the complete lack of track finding in #509 could potentially be fixed with a relatively small change.

[krasznaa]

Actually, never mind. Even in #509 some of the tests complain about a complete lack of track finding. 🤔

[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from KalmanFitTelescopeValidation2/KalmanFittingTelescopeTests
[ RUN      ] KalmanFitTelescopeValidation2/KalmanFittingTelescopeTests.Run/0
WARNING: mask store has empty collection no. 1
WARNING: mask store has empty collection no. 2
WARNING: mask store has empty collection no. 3
WARNING: mask store has empty collection no. 4
WARNING: mask store has empty collection no. 5
WARNING: mask store has empty collection no. 6
WARNING: mask store has empty collection no. 7
WARNING: material store has empty collection no. 0
WARNING: material store has empty collection no. 1
WARNING: material store has empty collection no. 2
WARNING: material store has empty collection no. 4
WARNING: acceleration data structures store has empty collection no. 1
WARNING: acceleration data structures store has empty collection no. 2
WARNING: acceleration data structures store has empty collection no. 3
WARNING: acceleration data structures store has empty collection no. 4
WARNING: No entries in volume finder
Detector check: OK
/__w/traccc/traccc/tests/cpu/test_kalman_fitter_telescope.cpp:131: Failure
Expected equality of these values:
  track_candidates.size()
    Which is: 0
  n_truth_tracks
    Which is: 100
[  FAILED  ] KalmanFitTelescopeValidation2/KalmanFittingTelescopeTests.Run/0, where GetParam() = ("telescope_100_GeV_0_phi_muon", { 0, 0, 0 }, { 0, 0, 0 }, { 100, 100 }, { 0, 0 }, { 0, 0 }, -1, 100, 100) (107 ms)
[----------] 1 test from KalmanFitTelescopeValidation2/KalmanFittingTelescopeTests (107 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (107 ms total)

Still, I can reasonably hope that whatever broke those tests, is what makes traccc_seq_example not find any tracks either. 🤔

[beomki-yeo]

@krasznaa I adjusted the input parameter to pass the CI test.

[krasznaa]

I'm generally on board, just some simple questions...

[beomki-yeo]

@krasznaa Just commenting on this change, the eta range is changing from [-1,1] -> [-0.3, 0.3].

Yeah so nothing has been fixed yet and I am avoiding to face the problem in this PR