Dear corry experts,
I came across the problem of track reconstruction with high particle density beams. Since the tracking4d module connects every cluster of the first and last plane, I was wondering if there is any method to try and distinguish between those clusters. I am aware of a chi2 cut in the alignment process, but is it possible to cut on chi2 in the final tracking, too? Or can this problem be somewhat solved with a different tracking module? I currently use the GBL track model and try to reconstruct protons simulated with allpix squared.
Thank you very much in adavance for your help
Christopher
Hi Christopher,
there is a trackChi2 cut on the final tracks.
However, it’s not part of the [Tracking4D] but you can apply it when using [AnalysisDUT], [AnalysisEfficiency] or any of the other analysis modules.
Cheers, Jens
Hi Jens,
thank you for your reply. In this case I only have residuals of the DUT with the Chi2 cut right? The residuals of the telescope stay the same, if I understand it right.
Cheers, Christopher
Hi,
I had one further question about tracking modules for high track densities. Are there currently track models, which work better under theses conditions, which are implemented in Corryvreckan, or could be implemented?
Cheers, Christopher
Hi Christopher,
I’m sorry for the late reply.
If you want to have a look at the telescope residuals after applying a track chi2, you can have a look at the module [AnalysisTelescope], this should give you exactly what you want:
And for your second question:
We’re supporting 3 track models currently:
- straight line
- GBL
- multiplet
I think generally straight lines work better at higher rates because of the smaller search cuts you can use when building track candidates. However, if a straightline track does not yield a satisfactory result (such as at the low energies at DESY), then there’s no way around using GBL and at high occuancies you will run into issues of combinatorics.
So I don’t think that any other (new) track model would help. The track model only describes the trajectory of one particle through the telescope.
What you mean is probably if the track finding (i.e. the finding of track candidates) can be improved. This is a question of how you combine individual clusters on all planes in to track candidates before “trying to fit them”.
Ideas how to improve Corryvreckan are always welcome 
Thank you for your Answer.
Yes, I meant to refer to the track finding. Do you know, whether someone already tried to create a track finding procedure suitable for high track densities? I try to look, if it is somehow doable for Corryvreckan to reconstruct tracks under these conditions.
For the best description of the track finding we currently have, you could have a look at Section 3.3 of this paper:
D. Dannheim et al 2021 JINST 16 P03008
https://iopscience.iop.org/article/10.1088/1748-0221/16/03/P03008
To my knowledge, currently nobody is working on improving this. Corryvreckan is always driven by the needs of its users, so as long as it works for what people need, it’s not developed further.
But I can encourage you to join our Mattermost channel and ask around there. Possibly, there is someone with a similar interest and you could team up and discuss:
https://mattermost.web.cern.ch/corryvreckan
P.S.: I hope, I have my notification settings right now, so the next reply won’t take so long
Hi,
I tried using the TrackingMultiplet Module, but a Segmentation Error ocurred after the Event Loop wants to start. I use Corryvreckan on Lxplus with the v1.0+1459^g5b47306 Version. Do you know what might be the cause of it?
Hi chkrause,
It seems that you’re using quite an old corry version (including some local changes ^g5b47306).
Maybe you can try to bump your version to the latest master (or v2.0.1) and see what happens then.
You could also try to run the test/example:
to see if your problem is related to your installation or to your configuration.
Cheers,
Jens
Hi,
thank you for your suggestions. I used the newest version now, but the errors stay the same, although they disappear, whenever I’m not using an AnalysisTelescope module afterwards. The test run however works and it also has an AnalysisTelescope Module.
Furthermore I am interested in using the Filewriter with the TrackingMultiplet data, and it goes through the event loop (while having no AnalysisTelescope Module active), but it basically gives me a warning, so I also wanted to ask, if you know something about that.
Okay, that’s good. So it means the issue must be somewhere in your config file.
Maybe you can start from the example and change it step-by-step to your configuration. Then you can see at which point it starts to fail.
As for the file writer and file reader, you can also check the examples.
For the warning, you may contact pschutze, who developed this module.
In the test examples the Eventloader Modules are used, while I use the FileReader to read in the data produced by Allpix. This is the only difference I can narrow it down to. Can that somehow cause problems?
Do you have the E-mail of Paul Schütze, because I did not manage to find one.
You can find his e-mail address in the README (or contact him via mattermost):
