@MasatoshiKobayashi @ramirezdiego - I ran some tests with these epix/wfsim updates, including a test with mc_chain (branch "lower_test"), simulating "InnerCryostat_Co60" with these branches. I've seen two issues so far:

• if we end up with zero events after applying the energy selection, the simulation will crash,
  either here:
  strax_interface.py#L823
  or here:
  common.py#L211
  for obvious reasons - the instructions are empty so we don't have these fields. We need to find a proper way to handle these exceptions.

• Since the selection is applied in epix, we still have an issue with large G4 files. For instance, in Co60 simulations we still cannot use more than 5E3 primaries per simulation or so, because the file size is already around 40 [Mb] with 5E3 primaries. It's a problem because (a) epix does not handle large G4 files very well right now since it does not read it in chunks; (b) we will still generate huge G4 output at the temporary directories at OSG. If we need 1E8 primaries for this single Co60 simulation, it's already close to 1 [Tb] of G4 output in total + we will need to run it with 2E4 jobs in parallel. We need to think whether we can apply some reasonable/safe pre-selection at G4 level.

Hi @PavelKavrigin , thank you for your check!
I am not so much familiar with the WFSim, but if possible the simplest way is to finish the epix if the selection is empty.

Regarding the G4sim, I'll have a look. I am apply pre-pre G4 level section, for example, if there is already the large E-dep in the TPC don't save the event.

@PavelKavrigin
If we just remove the step with position R>68cm or Z<-170cm, the size of G4 file could be reduced down to 1/3. Is it still possible to run the epix with this? Or do we have to keep all steps to run?

@MasatoshiKobayashi - I think it's OK if we select all steps within the TPC at the G4 stage. Energy selection is more tricky since we will need to sum up all energy depositions within the TPC, not within the whole LXe, and we also need to take into account the time of the interactions - if we do not postpone radioactive decays, for instance, we might run into problems if we simply reject all events with a total energy deposition in the TPC larger than a certain threshold.

@PavelKavrigin
I guess the simplest solution for the empty instruction is inserting something like:

if len(self.instructions_epix)==0:
  print("the instruction is empty")
  sys.exit(0)

around here? https://github.com/XENONnT/WFSim/blob/617532f7498af23e1bf808c8fd47df1477f82cf2/wfsim/strax_interface.py#L801

@PavelKavrigin
I pushed some lines so that to return the empty instructions if no events are left after the selection.
Also, wfsim side was updated to abort without crashing with empty epix instruction:
XENONnT/WFSim@473b314