-
Notifications
You must be signed in to change notification settings - Fork 157
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix: Quick fix LineSurface
tolerance inconsistency in globalToLocal
#2239
fix: Quick fix LineSurface
tolerance inconsistency in globalToLocal
#2239
Conversation
Codecov Report
@@ Coverage Diff @@
## main #2239 +/- ##
=======================================
Coverage 49.35% 49.35%
=======================================
Files 445 445
Lines 25264 25265 +1
Branches 11650 11651 +1
=======================================
+ Hits 12468 12469 +1
Misses 4512 4512
Partials 8284 8284
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
📊 Physics performance monitoring for b7a536eSummary VertexingSeedingCKFAmbiguity resolutionTruth tracking (Kalman Filter)Truth tracking (GSF) |
logged into this issue again and this is the output I get
you can see that both functions get the same input but conclude to a different distance. |
There is a sign flip in the momentum, this looks like it's the cause. |
Nope.
I don't think the direction should make a difference in this case and it seems to not do so. If I remember correctly there are a bunch of random sign flips in the navigation which might cause this inconsistency. But I don't think this is relevant now. |
ok, understood - probably time to have a look at the code lines. |
I agree but I wont have time for this in the near future. So I would propose to merge this quick fix and reference the underlying problem in an issue + in the code. |
linked #2226 in the description |
@andiwand thanks! |
updated the hashes @paulgessinger |
…l` (acts-project#2239) This is meant as a quick fix for the problem I am experiencing in acts-project#2086. It seems like it also appeared somewhere else already and maybe in one of the recent issues. The problem seems to be that the tolerance is not consistent between `intersection` and `globalToLocal`. Since these are used in sequence with the same tolerance `globalToLocal` will fail as it reports a higher distance than the `intersection`. A proper fix would be to align the math of the intersection and conversion so the tolerance has the same meaning in both cases. should also fix acts-project#2226
…roject#2086) Our full chain pulls are in a bad state. Looks like the reconstruction and simulation energy loss did not match up. This PR switches the Fatras interactions on which should bring our pulls back to standard normal distribution. Fixes - acts-project#1643 Blocked by - acts-project#2157 - acts-project#2239 - acts-project#2295 - acts-project#2293 - acts-project#2294
This is supposed to be a proper fix after the quick fix in #2239 In `globalToLocal` we did not respect the direction which is important for the line surface. I tried to make the code a little more clear and move away from constructor initialization in the mathy part. I also added a unit test to check the intersection against the propagation. I will let the CI judge how much this breaks
This is meant as a quick fix for the problem I am experiencing in #2086. It seems like it also appeared somewhere else already and maybe in one of the recent issues.
The problem seems to be that the tolerance is not consistent between
intersection
andglobalToLocal
. Since these are used in sequence with the same toleranceglobalToLocal
will fail as it reports a higher distance than theintersection
.A proper fix would be to align the math of the intersection and conversion so the tolerance has the same meaning in both cases.
should also fix #2226