Choose timezone
Your profile timezone:
CIDeR-ML General Meeting
→
America/Los_Angeles
Description
https://u-tokyo-ac-jp.zoom.us/j/83932834349
Link to Recording
Minutes:
Quick recap
The meeting focused on progress updates for ongoing research projects. SLAC reported that Section 2 of their paper draft is complete, with only the inference analysis and optimization comparisons remaining, targeting completion by the end of August. Riya presented work on laser calibration with LUCiD, showing that machine learning-based absorption field training can achieve results in about 5 minutes compared to 13 minutes for traditional Monte Carlo methods. The team discussed challenges with fitting parameters like scattering length and reflection values, with Kazuhiro raising questions about asymmetries in the peak distributions and suggesting the need to separate first and second peak photons for more detailed comparison. Patrick noted that differences between top and barrel rows could be due to geometry differences in PMT coverage.
Next steps
Patrick
- Ask Nahuel to look into the LUCiD code to extract true photon information for separating photons contributing to the first and second peaks in the laser calibration analysis.
Riya
- Provide Nahuel with Monte Carlo simulation data to use instead of the digitized histograms for the laser calibration project.
- Ensure that the physics in LUCiD matches SK's internal simulation (SKG4) as closely as possible for the laser calibration project.
SLAC
- Continue working on the inference analysis and optimization/comparison for section three of the paper, including running the simulator and Siren, aiming to finish by end of August.
Summary
SLAC Project Progress Update
SLAC provided an update on the project progress, noting that section two is complete with checkmarks added and chroma lookup table generations ready for integration. For section three, optimization and comparison of different ablation setups will require running the simulator and Siren, which will be done in the coming weeks. SLAC confirmed the end of August deadline remains achievable despite recent busy periods, stating that approaching deadlines typically motivate focused writing.
Laser Calibration Project Discussion
Riya presented Nahuel's final project on laser calibration using LUCiD to measure variations in water absorption, which showed promising results but had some discrepancies in parameter values like absorption length and refractive index. The team discussed using Monte Carlo simulation data instead of digitized plots to improve accuracy, with Patrick noting concerns about modeling the speed of light in water correctly. Riya agreed to use SKG4 to ensure physics in LUCiD matches SK simulation as closely as possible, as the next step to resolve the discrepancies.
Water Reflectivity Model Discussion
The team discussed issues with water reflectivity and Mie scattering in their model, with Riya explaining that Mie scattering is difficult to implement properly due to its size relative to the tank. Kazuhiro observed that the orange (paper) data shows a faster falloff in the second peak compared to the blue (fitted) data, particularly in rows three to five. Patrick suggested that the asymmetry in the top panel might be due to geometry differences between the top cap and barrel rows, as the top cap spans the entire horizontal space while each barrel row only covers a ring of PMTs. The team agreed to investigate further by separating first and second peak photons using truth label information to better understand the physical differences between reflection and direct photon detection.
