Choose timezone
Your profile timezone:
CIDeR-ML General Meeting
→
America/Los_Angeles
Description
https://u-tokyo-ac-jp.zoom.us/j/83932834349
Recording
Minutes:
Quick recap
The team discussed ongoing modeling issues related to photon generation and PMTs, with Junjie explaining the necessary fixes and Ka Ming providing guidance on file formats and conversion scripts. Progress updates were shared on various projects including LED calibration, SIREN code learning, and Omar reporting on a finalizing paper and Cesar mentioning code debugging. The team coordinated upcoming meetings and presentations, including the Hyper-K meeting next week and WCTE following, and agreed to cancel their regular meeting due to scheduling conflicts.
Next steps
- Ka Ming: Share scripts used to generate photon data files (for calibration/photon shotgun/diffuser ball studies) with Ryotaro.
- Riya: Set up IPMU cluster to run the SIREN (and LUCiD) code.
- Riya and Vera: Meet up today to discuss the SIREN code.
- Omar: Rerun all plots for the LUCiD paper and aim to upload a revised version to the archive by early next week.
- Ka Ming: Reference the Lucid paper in the upcoming Hyper-K meeting on Tuesday (contingent on paper being available on archive).
- Patrick: Review paper on Tuesday.
- Omar: Collect comments from Patrick and others on the LUCiD paper for possible revision before submission to PRX.
Summary
Modeling Error Correction Discussion
The team discussed an issue with high-visibility regions in their modeling, which was caused by a modeling error where photons were incorrectly being generated inside PMTs in a vacuum environment. Junjie explained that they are fixing this error by generating a new photon library and restarting training. Ka Ming mentioned he was busy preparing for HyperK and WCTE meetings and could not work on CIDeR this week.
MC Sample and LED Calibration
Riya had no updates to share. Ryotaro successfully generated an MC sample and is working on LED calibration, and did not know Ryo's status.
WCSim to H5 Format Conversion
The team discussed the existing pipeline for converting WCSim to H5 format for CIDeR input, with Ka Ming confirming that H5 is a generic format and can provide scripts used in previous simulations. Patrick emphasized the importance of using a single tool to maintain consistency across all simulations (and save Ryotaro's time), and Ryotaro acknowledged the need to check the accuracy of conversions. Ka Ming mentioned that calibration data needs to include intensity information as well, which would require additional columns compared to existing formats like photon shotgun or PMT hit only data.
SIREN and LUCiD Updates
Vera and Riya will meet today to get the code running. Ka Ming inquired about running the code on the IPMU cluster, to which Riya responded that she needed to set it up. Patrick asked about progress on the LUCiD code, to which Riya replied that there was no progress yet.
Paper Finalization and Hyper-K Meeting
Omar reported that a paper is being finalized, with Cesar expecting to upload it to the archive by next week after fixing a bug in the code that improved reconstruction results. Ka Ming mentioned he wants to reference the paper in the Hyper-K meeting next week Tuesday, but Patrick noted he was planning to review the paper first on Tuesday, so they may need to upload another revision later.
Omar indicated that most main items were in place and suggesting that comments from team members would be needed for final adjustments, potentially avoiding a second review. He confirmed the target channel as PRX.
Patrick proposed to cancel next and next next week's meetings due to the Hyper-K and WCTE meetings, with the possibility of a separate US-side meetings if needed.
