CIDeR-ML General Meeting

→ America/Los_Angeles
Description

https://u-tokyo-ac-jp.zoom.us/j/83932834349

Recording

Minutes:

Quick recap

The team discussed visibility map analysis and simulation results, focusing on patterns around PMTs and potential issues with discontinuities in the detector medium. They explored various technical challenges including Siren's performance near boundaries, model sampling details, and calibration problems with LED MC systems. The conversation ended with work transitions between team members and plans for future discussions on specific technical issues.

Next steps

  • Sam: Run a small simulation study to characterize the visibility variation as a function of distance from a photon to a single PMT (within 100 millimeters) and investigate the angular response, to better understand the behavior near the PMT.
  • Junjie and Sam: Check the scattering parameter settings in Chroma and investigate whether observed features (e.g., secondary walls) correspond to scattering effects.
  • Ka: Support the transition of responsibilities to Rhea and Vera before leaving next month, ensuring all relevant tasks and information are handed over.
  • Ka: Send Ryotaro the command/configuration needed to address the LED MC output issue and discuss offline as needed.
  • Ryotaro: Check and resolve the LED MC configuration issue (with support from Ka) and follow up with Ka as needed.

Summary

Orthogonal Bands in Visibility Maps

Junjie presented an analysis of orthogonal bands in visibility maps, showing how photon positions and PMT detections correspond to visibility points. He explained that high visibility values occur near PMTs, with interesting patterns forming rings around PMTs when predictions exceed truth, and positions directly on PMT centers when predictions are lower. Junjie suggested running simulations of single PMTs to better understand visibility variations close to PMTs and angular responses, as he suspected scattering might cause fluctuations in the visibility map.

PMT Discontinuity Analysis Discussion

Sam and Junjie discussed discontinuities in the space between the detector medium, glass of the PMT, and photocathode, which they agreed to study further. Junjie noted some circular features in other bands and suggested examining scattering parameters in the Chroma. Patrick inquired about the simulation's geometric detail, and Sam explained that PMT models were based on Hamamatsu data sheets, with wire planes solved analytically for precision.

Siren Performance and Boundary Issues

The team discussed issues with Siren's performance near boundaries, observing diagonal bands in simulation results that were not present in previous work by Kavan. Junjie and Sam noted that their current simulation uses more detailed sampling and could be more susceptible to overfitting, suggesting the need to increase model expressivity or add damping functions. Ka mentioned a related but different issue with underestimation of visibility near PMTs due to insufficient pre-training samples, and agreed that tuning Siren's internal frequency terms could help address these problems.

Model Grid and Photon Distribution

Patrick and Junjie discussed the grid structure and empty spaces in a model, with Junjie explaining that the empty spaces are likely PMT grids where photons should fill the volume. Sam clarified that the union of the distributions in the bottom left and right should cover the entire distribution of positions.

Project Transitions and Work Plans

Riya and Vera discussed their work plans, with Riya focusing on completing the Lucid project and Vera taking over Ka Ming's work on Siren before joining Lucid. Ka mentioned supporting the transition work for bare area before leaving next month. Patrick encouraged quick verbal reports and ended the conversation without any additional questions.

LED Calibration and Digital Heat Issues

Ryotaro discussed calibration issues with LED MC and resistance output, noting that the log showed no errors but the digital heat feature was not functioning as expected. Ka suggested that Ryotaro might need to change a configuration setting and offered to send a relevant command, with both agreeing to discuss the matter offline later. Sam had no updates to report, and the conversation ended with a plan to hold another brief meeting the following week.