add routines for computing time coincidence map and evt

[tdixon97]

This is still a bit WIP.

• builds a simple TCM
• computes events using pygama functions

Issues:

• building the TCM is done fully in memory (potentially bad?)
• no tests yet
• some more flexibility is needed, eg. parsing expressions for channel lists etc.

I still think its useful if @gipert can have a look so we plan what needs to be changed

[codecov]

Codecov Report

Attention: Patch coverage is 3.65854% with 79 lines in your changes missing coverage. Please review.

Project coverage is 69.03%. Comparing base (f0f1791) to head (edabe63).

Files with missing lines	Patch %	Lines
src/reboost/build_evt.py	0.00%	50 Missing ⚠️
src/reboost/build_tcm.py	0.00%	29 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #51      +/-   ##
==========================================
- Coverage   72.30%   69.03%   -3.27%     
==========================================
  Files          23       24       +1     
  Lines        1607     1686      +79     
==========================================
+ Hits         1162     1164       +2     
- Misses        445      522      +77     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[gipert]

please wrap!

[gipert]

Can you calculate how many events you would have to process to make memory usage problematic?

[tdixon97]

I think around 10^8 hits over the full detector.

• I.e need to read one int and one double per hit, so 1 GB of memory would be 0.8 x 10^8 hits.
• There are probably cases we will run into issues, eg a 2vbb sim will be few 10^8 hits.
• In general run partitioning will help this a bit, but I rather not rely on it.

Note: The issue is just when we create the TCM, once we have it we can iterate over it without memory issues.

[tdixon97]

I think this can likely be done without loading the full file into memory with a "leap-frog" like reading. Similar to the build_glm code?

reboost/src/reboost/build_tcm.py

Line 39 in edabe63

hash_func = r"\d+"

[tdixon97]

Alternatively, build_hit can produce a hlm object telling which rows in the hit table correspond to which g4_evtid to help with iterating over the hit files... I am not sure about producing too many new objects though

[tdixon97]

I think we should merge this, since it is still useful functionality, even if some parts have to be changed to work for very large files