multiple actions within same subdetector

[danieljeans]

we're simulating a calorimeter with a mix of scintillator and cerenkov layers.
For the scint layers, we'd like to use the usual default calorimeter action, while for the cerenkov ones we'd like to use a new user-defined action.
what's the best way to implement this? should we define a single new action which can differentiate between C and S layers and apply the appropriate method, or is there some smarter way to do it, e.g. based on the "slice" index within a layer, which could avoid the need to duplicate the scintillator part?

a second question:
in ddsim, we can control which action to use using e.g. in the case of ILD, we have
SIM.action.mapActions['tpc'] = "TPCSDAction"
what exactly is the string 'tpc' matched with?
in ILD/compact/ILD_common_v02/tpc10_01.xml
we define

    <detector name="TPC" type="TPC10" vis="ILD_TPCVis2" id="ILDDetID_TPC" limits="Tracker_limits" readout="TPCCollection" insideTrackingVolume="true">

my guess is that's it a case-insensitive match to the ' name="TPC" ' ; if you could confirm or correct, that would be helpful.

many thanks,
Daniel.

[atolosadelgado]

we're simulating a calorimeter with a mix of scintillator and cerenkov layers. For the scint layers, we'd like to use the usual default calorimeter action, while for the cerenkov ones we'd like to use a new user-defined action. what's the best way to implement this? should we define a single new action which can differentiate between C and S layers and apply the appropriate method, or is there some smarter way to do it, e.g. based on the "slice" index within a layer, which could avoid the need to duplicate the scintillator part?

a second question: in ddsim, we can control which action to use using e.g. in the case of ILD, we have SIM.action.mapActions['tpc'] = "TPCSDAction" what exactly is the string 'tpc' matched with? in ILD/compact/ILD_common_v02/tpc10_01.xml we define

    <detector name="TPC" type="TPC10" vis="ILD_TPCVis2" id="ILDDetID_TPC" limits="Tracker_limits" readout="TPCCollection" insideTrackingVolume="true">

my guess is that's it a case-insensitive match to the ' name="TPC" ' ; if you could confirm or correct, that would be helpful.

many thanks, Daniel.

Hi Daniel,

A straightforward approach that requires no additional code is to use an optical tracker sensitive detector action. This action records photon information (time, energy, cellID, etc) as they enter the sensitive volume and immediately kills the photons. While Geant4 handles optical photon generation, solving a photon transport equation can be much faster than Geant4 tracking. I guess that by encoding into the segmentation whether the layer is C or S, the corresponding signal can then be reconstructed.

Perhaps you have already considered this approach and found reasons not to use it. Regardless, I’d be interested in learning what you think is the best option for your needs! :)

I had the same question about the map of actions when I started with DD4hep :)
The key of the map is the name of the detector. To avoid further confusions I usually append "_T" (for detector type) to the name of the detector constructor., eg here

[andresailer]

Question 2, as also stated by Alvaro:

DD4hep/DDG4/python/DDSim/Helper/Action.py

Lines 23 to 26 in abb85cb

	For specific subdetectors specific sensitive detectors can be set based on patterns in the name of the subdetector.
	>>> SIM = DD4hepSimulation()
	>>> SIM.action.mapActions['tpc'] = "TPCSDAction"

Question 1:

I think this should be possible with multiple readouts, as there is an example here

DD4hep/examples/CLICSiD/scripts/SiD_ECAL_Parallel_Readout.py

Line 54 in abb85cb

# Setup the parallel readout for the detector EcalBarrel

And I wanted to implement this for ddsim, but I am not sure this will happen this year.

[atolosadelgado]

It is also possible that one sensitive detector action creates 2 output collections with one readout, right?

[MarkusFrankATcernch]

Yes. This should be possible. See here for an example:

https://github.com/AIDASoft/DD4hep/blob/master/examples/ClientTests/compact/MultiCollections.xml

[danieljeans]

thanks to all for your useful inputs!

regarding the action for the cerenkov sensitive detector, we are currently using a simplified approach, taking the sum of step lengths of charged particles with v>c/n within a cell as an ansatz for the cerenkov signal, to be converted to a sipm signal in a later digitisation step.
I guess that this should be much faster than doing the proper optical treatment? (I have not tested this assumption, and am open to being corrected!)