Choose timezone
Your profile timezone:
EP R&D Software Working Group Meeting
→
Europe/Zurich
32/1-A24 (CERN)
EP R&D Software Working Group Meeting
Zoom Meeting ID
62893115044
Host
Graeme A Stewart
Alternative host
Andre Sailer
Useful links
Join via phone
Zoom URL
---
tags: eprdet, wp7, minutes
---
Software R&D Working Meeting Minutes
====================================
## 2025-09-17
Room: Andre, Vincenzo, Alberto, Juan, Mateusz, Severin, Witek, Lukas, Dev, Aaron, Giacomo, Peter,
Remote: Anna
Apologies: Andi, Aurora, Florine
## News
### EP R&D Seminars
We were requested to give a presentation at one of the next EP R&D seminars https://indico.cern.ch/category/13619/. These presentations are meant for a general audience, so more context/motivation about the work presented should be given.
Giacomo: I can give my presentation about RNTuple attributes and extend it to fit this series.
Severin: in principle we could also give a presentation, but I would say in early 2026.
## Turnkey Software Stack
[Severin]: All across Key4HEP you use the same classes e.g. for a Calorimeter hit? How does it work for simulation? All projects using Key4HEP have to comply with this EDM? e.g. in CMSSW they do fancy things for calorimeters.
[Juan]: If they want to use EDM4HEP they have to comply with the format. For example in simulation you would have a specific EDM4HEP writer which feeds all calorimeter hit information.
[Severin]: Is what calorimeter hit has sufficient or do people put their own stuff on top of it?
[Andre]: Once in a while people make request to change the EDM, and often they don't actually need those things.
[Severin]: But do all detectors do different things on top or do they use EDM4HEP vanilla? Mostly I'm asking because at ACAT they asked me if we could run Adept with EDM4HEP.
[Andre]: If there's the need we can extend EDM4HEP, but only when it's needed. If it's just a matter of better documenting data model so people know what they can use, then we'll do that instead.
[Severin]: for us it would be nice because we have to integrate with different experiment frameworks, but if we integrate with EDM4HEP it could be one format for all. Which one should be the model used in the future?
[Andre] I hope they use DDSIM and DD4HEP.
[Juan]: We also support "extensions" to create your own data type before adding to EDM4HEP.
[Severin]: Is DDSIM based on Gaudi?
[Juan]: No, it's completely separate. You run your sim independently, you get an output file, then run your algos with Gaudi.
[Severin]: Why not all in one pipeline?
[Andre]: Some people stayed with DDSIM coming from the linear collider community? There's the desire to have everything in one pipeline but needs work and is non trivial.
[Severin]: Do you have a full geometry model implemented for your tests or do you specify a geometry and then gets translated to Geant4?
[Juan]: The geometry is DD4HEP, then we implement it in k4geo. If you want to run also simulations then you transform to Geant4
[Andre]: We read XML files with parameters which are converted to TGeo, then Geant4 is an optional step.
[Andre]: About the slower performance, there is no view in the HSF repository.
[Severin]: So what's the future for Key4HEP?
[Andre]: If funding does not continue, people will have problems. Key4HEP was mentioned in the summary slides of the Symposium in Venice.
[Severin]: Are users happy?
[Juan]: In general everyone finds a way to do their work, we can't always make everyone happy but I think we manage.
[Severin]: How much more are the software frameworks by experiments building on top of Key4HEP?
[Juan]: They use the same framework. DDSIM for simulation, Key4HEP for the data and analysis. What was called FCCSW is now Key4HEP. What is different for each detector is the algorithms for different trackers/calorimeters depending on the experiment. They use DD4SIM, and the rest is common. We are trying to have a similar validation chain, each detector provides a different script to run their own validation/RECO and we produce the plots. Material scans are all in k4geo, we can run them easily from there.
[Mateusz]: They also have analysis frameworks built on top of Key4HEP right?
[Juan]: Yes, I didn't cover it because it's difficult to define the boundary between Key4HEP and frameworks using it
[Severin]: I remember a new proposal for an asymmetric collider. I remember someone back then claiming it was a problem for DD4HEP. Are you aware?
[Andre]: I heard of HALF, they never reached out. There's some flags that say "replicate the same calorimeter on the other side", but we can always build one front and one back. LHCb is using DD4HEP so it should work.
[Vincenzo]: I was curious about schema evolution: how does that work in the case of Key4HEP/PODIO were multiple data formats are supported?
[Juan and Mateusz]: When writing to ROOT formats we can use the ROOT schema evolution, but we have different syntax/API to provide schema evolution support to other outputs.
## AOB
There are minutes attached to this event.
Show them.
