SiPM dsp parameters

[rosannadeckert]

Now we have the following parameters: blfc, timestamp, eventID_fadc, e_fc, trig_pos, trig_max, trig_pos_DC, trig_max_DC. Can we add the following parameters:

• tp_max, tp_min, wf_max, wf_min (finds max and min of waveform with corresponding time points)
• tp_max_lar, tp_min_lar, wf_max_lar, wf_min_lar (finds max and min of waveform with corresponding time points in a LAr scintillation window around 48 us, i.e. from 47 us to 53 us)
• wf_fwhm (FWHM of waveform projected on y axis)
• is_valid_hit (for each trig_pos, True or False if there was a crosstalk (DC) event close)

Let me know what you think.

[oschulz]

Sure, why not.

What do we use wf_fwhm for?

[rosannadeckert]

sometimes there are "noise trains" in the traces that we can tag with wf_fwhm

[oschulz]

Hm, we'll have to see how to make this performant. Variance would be easy, but for FWHM we have to build per-waveform histograms, right?

[theHenks]

So the max pars are no problemo. That is just adding something. The wf_fwhm could be a bit harder since you have to build histograms. I wouldn't recommend that since it will really take a lot of time. We could just get some signalstats from some part of the waveform if that is fine with you? Or we really calculate the sigma of the distribution. I can think about something. The is_valid_hit I would add in the event building stage.

[rosannadeckert]

that's how the build the histogram in pygama: https://github.com/legend-exp/dspeed/blob/main/src/dspeed/processors/histogram.py. Maybe we can do something similar. I agree that is_valid_hit can go in the evt tier.

[oschulz]

that's how the build the histogram in pygam

We should be able to use StatsBase.Histogram or FHist.Hist1D without high mem-alloc cost if we preallocate the hist weights as an ArrayOfSimilarArray for all waveforms, and then just wrap histogram objects around them.