Participants: J. de Cuveland, V. Friese, N. Herrmann, D. Hutter, G. Korcyl, P.-A. Loizeau, D. Smith, B. Sobol, A. Toia

● mCBM beamtime February 2025

• Observation: the CPU load was rather low (less than 10%), although tracking was run. The reason is unknown; possibly, multi-threading was not properly enabled. To be investigated. In general, we need better understanding and control over this.
• The BMON showed problems (quick degradation). In HADES, the BMON is re-calibrated each 15- 20 minutes.
• Analysis of the taken data should use the binary offline. The knowledge on how to do this has to be spread.

● Data challenge / Vertical test

The current online process cannot yet consume DigiTimeslices as input, skipping the unpacking stage. This can be achieved, however, without much effort.

A replay mode for simulated data (DigiTimeslice) is not easily possibly with the current replay scheme, which is tailored to replay FLES timeslices. To use the same scheme for DigiTimeslice data would require major investigations and work. On the other hand, just reading and processing data from files has limited learning potential. The most appropriate is probably to install a separate replay scheme using e.g., FairMQ.

Whether mis-alignment (or other mis-calbrations) will be introduced in the Vertical Test is not yet decided. Although these are important issues, the danger is to overfreight the preparations. An option is to introduce these additional complexities in a follow-up VT.

● Development goals and priorities

• S. Zharko and Souvik Chattopadhyay: Analysis of mCBM data. Compare triggered (V0 topology trigger / KFPF Lambda trigger) to minimum-bias events.
• A. Senger: Trigger study for Vertical Test
• V. Friese: Enable binary to consume DigiEvents generated from simulation
• B. Sobol: Look into the multi-threading issue.
• Comparison ROOT and BOOST streamer: the thesis in Frankfurt investigating this issue is completed, but the results are not known yet. The relevance of this has diminuished, though, since the new ROOT data model RNTuple is already available (with FairSoft Jan24).