Skip a CKF when no measurement is found on a surface #529
Conversation
|
Looks good |
|
@paulgessinger Can you guess why the CI Bridge was refused? |
|
That is only enabled for a selective list of accounts. So this is expected. We can add her later once/if PRs become a regular occurrence. 😉 |
|
I think she needs to be added to the list because this PR is definitely affecting GPU codes as well |
|
@krasznaa @paulgessinger Could you guys consider adding her to the CI account list? I am aware that the GPU-CI is not a free machine and it belongs to a certain group. But Shima has contributed to this project already a couple of times, and it is going to be easier to maintain our project in a safe manner with her future contributions |
|
I sent the invite |
|
Thank you so much! |
|
It seems the CUDA CI fails: log I think it is not a problem of this PR but something else |
There was a problem hiding this comment.
I think the GPU CI is not working very well. Let's just go with it. @shimasnd could you rebase the PR so it can get merged?
krasznaa
force-pushed
the
skip-a-CKF-step
branch
from
d33cf47
to
43140e9
Compare
It is to continue CKF steps even if there is no measurement on an expected surface. (In current traccc, CKF simply stops for a candidate track in such a case and the candidate is discarded.) As a tentative solution, a dummy measurement is inserted to extend CKF steps.
Additional modification is to keep track candidates when the number of CKF steps reaches the max_track_candidates_per_track.