general:
- discussed mail sent by Oliver about the usage of anaconda. Some of us use anaconda on maxwell. For us it was not clear, whether the problem with the licence also applies to the version of anaconda installed on maxwell or only to locally installed versions of anaconda. Engin suggested to contact someone responsible for maxwell.
Engin:
- working on training the ECAL+HCAL generator at the same time; so far the training does not give good results yet
- looked into libtorch in Geant4, wants to do the timing study Francisca did for her thesis again, since this has to be updated
Peter:
- is preparing his talk at GSI and is writing proceedings for ICHEP
- worked on energy rescaling approach for the energy conditioning; fixed bug that fixes the cell energy; Engin suggested to run the reconstruction with the newly trained samples; Peter and Engin will look into this together on Friday
Sascha:
- was on vacation, doing administrative stuff like introducing new students to the group, working on PhD defence and applications
Anatolii:
- is writing report about his PhD studies
- attended Terascale school on ML
- continues to work on 1D showers
Lennart:
- re-ran model; now the best model gives similar results for the validation and test data (had class imbalance, now weighting implemented to account for class imbalance)
- first draft of paper will be circulated on Friday
Mareike:
- presented work in iLCSoft and Analysis meeting on Monday; got a few more suggestions what to implement and check; will give a follow-up presentation in the next meeting on Monday
- reading about architecture used by CMS to understand in detail what they used and why
- question by Engin: Will you use exactly the same model used by CMS? Is the CMS architecture public somewhere? answer: will start with the architecture used by CMS, adapt it to our needs, so far, not found; Engin suggested to contact the authors and ask for it
Birgit:
- long-standing issue, that if you try to generate Delphes events with Pythia for resonances, this does not work; met people at ECFA workshop and discussed this; hopes that this will get fixed now
Finn:
- worked on source code for k4LCIO Reader and the converter
Christiane:
- looking into GraphNNs for track reconstruction, looks promising, currently looking into newer studies where the track density is higher; looking into the details to see what was done
Mareike:
- will give FTX-SFT summary in the next FTX group meeting on Tuesday; discussed the highlight; nobody had an idea for the highlight (last time we had the radio galaxy studies from Lennart)