Specific documentation on ROOT for traccc

[SylvainJoube]

Added a small ROOT documentation for traccc, as discussed on the Friday Feb 23, 2024 Acts Parallelization Meeting.

[beomki-yeo]

Can we focus on the interpretation rather than providing exhaustive tutorials on how to use ROOT?

It will be very useful to explain how the tracking efficiency and duplicate rate are calculated. Of course you are providing simple definitions but I believe there are more rooms to be clarified. For example, you can explain how to decide whether a certain pattern is matched with a truth particle. (e.g. 100 % match in the pattern? or 50 % ?)

You can also add exaplanations on track fitting result (e.g. How the pull values are calculated)

[beomki-yeo]

I think we can skip this

[beomki-yeo]

We can just forward people to the official link: https://root.cern/install/

[SylvainJoube]

I updated the documentation, I understand that adding our own documentation on how to install ROOT is one more component to maintain, and should be left to the ROOT maintainers.

[beomki-yeo]

This can be skipped as well because it is already described in the main README

[SylvainJoube]

Got it, I replaced those lines by saying 1) to compile preferably ROOT and traccc with the same compiler to avoid linking errors and 2) not to forget to source the ROOT script before the execution of traccc.

[SylvainJoube]

Thanks for your feedback Beom Ki. I will try to explain how the tracking efficiency and duplicate rate are calculated and how the track fitting ROOT output is designed.

[stephenswat]

Very minor, but some of your file names say duplucation which should be duplication.

[SylvainJoube]

Thanks Stephen! I missed that one! 😹

[SylvainJoube]

A few bottles of milk later... I'm stuck, i don't know how to download / generate the "-hits.csv" files.

The bin/traccc_seq_example works fine on my machine, as long as I don't append the --check-performance flag to the command line. Adding the flag randomly creates either :

WARNING: No material in detector
WARNING: No entries in volume finder
Detector check: OK
terminate called after throwing an instance of 'std::runtime_error'
  what():  Could not open file '.../traccc/data/odd/muon100GeV-geant4/event000000000-hits.csv'
Aborted (core dumped)

or a *** Break *** segmentation violation

I think the event00000000X-hits.csv files are missing from the archive. I also tried with traccc-data-v6.tar.gz and v4 but without success.

[krasznaa]

Unfortunately this is something that we discovered as well. 🤔 The current ODD files we have in in our TGZ files do not have all the necessary information for making physics performance plots. 😦

I now became a little unsure, but I believe we discussed this with @asalzburger last Monday. That he (or possibly @beomki-yeo?) would make us some new simulated files soon. Since we'll very much need those for the measurements that I'll want to have in 2 weeks from now. 😉

[beomki-yeo]

I thought it was supposed to be Andi. I am not familiar with the ACTS core simulation

[beomki-yeo]

As a note, -hits.csv is not necessary for seq_example (we need all other additional files though) -- somehow the check-performance is designed to require -hit.csv but this will need to be fixed. Let's hope Andi hands us over the hits.csv as well so we don't have to deal with this right now :p

[SylvainJoube]

Thanks a lot for your replies Attila and Beom Ki!

I will use the seeding_example to finalize this documentation then, with the data generated by the toy_detector. If I'm not mistaken, don't think I can use the seq_example with the toy_detector data, since it generates measurements.csv, measurement-simhit-map.csv, particles.csv, hits.csv, but we also need cells.csv for the seq_example full chain.

[SylvainJoube]

Sorry for the wait, I’m mostly contributing to traccc in my free time these days, I have a lot to do on other fronts! 🫠 I have finally added documentation for the duplication and efficiency plots for the finding and ambiguity resolution algorithms. I will ask Corentin for help for the fitting algorithm plot documentation, but maybe, if you judge it appropriate, we could merge this PR and open a new one for the rest of the documentation?

[SylvainJoube]

Sadly, I don't think I'll have time to add a substantial documentation for the fitting algorithm in a few weeks, is there anything you would like me to change in this PR @beomki-yeo ?

I think this whole PR may became irrelevant in a few weeks due to Attila's work on the new data format for ROOT plots, so depending on you thoughts, we may merge or close this PR.

[beomki-yeo]

Can we change the directory name or document title? The contents seem to be about how to generate and interpret the performance plot, instead of how to build and use root which should not be introduced in traccc

Maybe "root -> tracking_performance" and "Specific documentation on how to build and use ROOT for traccc -> "Specific documentation on how to generate and interpret tracking performance plot"

In main README.md, there is also a very brief explanation on performance plot. I would add a link which forwards to this doc

[beomki-yeo]

And the size of images is way too big -- I suggest to shrink them

[stephenswat]

And the size of images is way too big -- I suggest to shrink them

Hi @beomki-yeo, could you please clarify if you mean the file size is too big, or if the resolution of the images are too high?

[beomki-yeo]

I meant the spatial size :D

[stephenswat]

This should say TBrowser.