EDM4HEP Discussion #3

1/1-025 (CERN)



Show room on map
Videoconference Rooms
Vidyo connection for EDM4HEP discussions
Graeme A Stewart
Auto-join URL
Useful links
Phone numbers

EDM4HEP Live Notes

This is a document for taking notes during EDM4HEP meetings.

Present: Frank, André, Valentin, Graeme, Geri, Xiaomei, Weidong, Jiaheng


Tuesday 11 seems like a good time for most people. We will propose that.

Framework WG should look at lower level “common” tools that would be of interest (cf. concurrency forum).

EDM4HEP Progress and Discussion

MCParticle has a lot of extra helper functions, which we believe to be very useful.

A lot of LCIO thinking/discussion went into the information needed (represented in the bitfields). Convenience functions are good to give to users - make the representation of the data an internal property of the EDM (ask people not to use it directly, though it does need to be in EDM).

Hadron case: pile-up and beam background would be useful to have.

Code for mother/daughter requires book-keeping of maps, which is a bit awkward.

PR for hits discussed. LCIO streamers encoded a lot of extra information (details of G4). Not so easy in PODIO. Thus implemented as two objects, which is not very elegant, but didn’t find another way.

SimCaloHit contains position - there are a lot of these so is it too much to have this for every hit? In LCIO the position was optional, so it could be left out as needed (supported better in DD4hep now).

Question - can we have optional information in PODIO? Not trivial with PODs.

FCC used multiple objects and composition (barehit, etc.). Valentin needs to check on this - TODO for the next discussion.

Proposal - accept the PR to allow people to code against it and play with real use cases.

Should cellID be connected to the DD4hep readout string definition? Of should DD4hep make it easier to rettrieve this information? Add something and then play :-)
As an example: The CellId definition for a Hit could look like <id>system:4,layer:5,module:18,x:-15,z:-15</id>
Saving this alongside the hit will make using the cellId easier.

Agree to merge current PRs to help the bootstrap

System of Units

For energy we’ll use GeV. This should be documented. (Add it as some metadata?)

MCParticle creation time is ns.

Distance in mm.

If you want different units create a different object.

Ship a units file with EDM4HEP? It’s a very rare use case and there seem to be too many of these already. G4Units.

CI using latest PODIO ?

Could build PODIO on the fly? Yes, can be the part of the way the CI would be configured and go into a matrix of tests.

CVMFS - try to get /cvmfs/hsf.org/sw blessed for the wider world (ACTION on Graeme and Jakob).

Example code/ Tests

  • examples for writing and reading
  • can also serve as tests
  • already have HepMCtoMCParticle

can we store meta data with the collections ?-

  • would like to have named parameters a la LCIO
  • types: string, int, float (double) and vectors of these …


Happy so far with what we have. (Don’t hesitate to raise issues that you see.)

There are minutes attached to this event. Show them.