Add corryvreckan converter script

[cbespin]

Adds a standalone script to convert interpreted data to a format compatible with the corryvreckan EventLoaderHDF5. Supports:

• AIDA mode (timestamp-based, "forwarding" hits to the EventLoader)
• legacy EUDET mode (pre-built events from DAQ converted to the EventLoaderHDF5 file format).

@maxbab1 Please have a look if you spot any mistakes

[maxbab1]

Thanks for the script!

I was not able to testshi script, because after interpretation, i don't have a table 'Hits'. I only have a table 'Dut'.

Also the Dut table does not have the TLU words associated with the hits yet. I had to do this by meself in my QnD converter script.

[cbespin]

The DUT table is everything that is needed if trigger_mode="AIDA" since we do not rely on any events that the DAQ software produces. In case of trigger_mode="EUDET" we need the data that has been created wit the event building functionality in the DAQ software since the timestamps are not synchronized with other devices in the beam but only locally between TLU and DUT words. The association is done with the internal event builder.

[maxbab1]

Works in the eudet mode with my custom event builder.

However when i try to use the event builder from the repo, there is an issue:

2024-12-17 12:13:29,053 [Analysis         ] - ERROR   Exception during analysis
Traceback (most recent call last):
  File "/mnt/Daten/Nosync/Schule/tj-monopix2-daq/tjmonopix2/analysis/corryvreckan_converter.py", line 93, in <module>
    a.analyze_data()
  File "/mnt/Daten/Nosync/Schule/tj-monopix2-daq/tjmonopix2/analysis/analysis.py", line 379, in analyze_data
    hist_cs_tot += cs_tot.astype(np.uint32)
ValueError: operands could not be broadcast together with shapes (256,) (512,) (256,) 
Traceback (most recent call last):
  File "/mnt/Daten/Nosync/Schule/tj-monopix2-daq/tjmonopix2/analysis/corryvreckan_converter.py", line 93, in <module>
    a.analyze_data()
  File "/mnt/Daten/Nosync/Schule/tj-monopix2-daq/tjmonopix2/analysis/analysis.py", line 379, in analyze_data
    hist_cs_tot += cs_tot.astype(np.uint32)
ValueError: operands could not be broadcast together with shapes (256,) (512,) (256,) 
  0%|                     

I am not 100% sure if the repo event builder fails because my data format is different or if this is a general issue and noone used it so far...

[cbespin]

The feature has been used / is used all the time, it originates from the clusterizer that is integrated for standalone analysis. What is your TOT binning? Did you apply a charge conversion already at some point? Otherwise it might be your data format

[maxbab1]

If i use my custom event builder it works fine.

But if i just let the script interpret the h5 file we have, it fails. But my data file uses the TLU-format with only trigger number, so i guess that's why it is failing. the results would be useless anyway and that's the reason i have my custom event bulder.

[cbespin]

I guess that will be the case since the clusterizer works on event basis in this case which is not really possible with the data format you have available. I will fix the column and row numbering, would you approve the PR then?

[maxbab1]

All good now! Thanks!