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 updates and discussions related to the PMT angular response and optical simulation models in the context of Super-Kamiokande data analysis. Zhenxiong presented work on extracting PMT angular response using SIREN with WCTE cosmic data, comparing pre-tuned and post-tuned models, and discussed differences observed between PMT groups, possibly due to geometric effects. Riya shared progress on updating physics models in Lucid to align with SK's internal simulation, including PMT and black sheet reflectivity, and compared time-of-flight distributions between Lucid and SKDetsim, noting improvements and remaining discrepancies. The group discussed potential sources of differences in the timing distributions, such as PMT geometry, absorption, and electronics effects. Riya also presented initial fitting results for optical parameters like Rayleigh and Mie scattering lengths, showing reasonable agreement with injected values. Ryotaro briefly mentioned ongoing work on LED and pion studies. The conversation ended with suggestions for further investigation into model implementations and comparisons.
Next steps
Riya
- Make a 3D loss landscape plot for the joint fit of Rayleigh scattering, Mie scattering, and absorption lengths.
- Remake the 2D loss landscape plot with a focused range and color scale to better show the minimum.
- Look into finer binning for the MC time-of-flight histograms to better analyze the peaks and offsets.
- Document the physics models being used in Lucid.
- Implement Nahuel's SIREN method for top-bottom absorption asymmetry correction in Lucid.
- Implement Tituan's SIREN for PMT QE corrections in Lucid.
- Try simultaneous fitting of more parameters (e.g., black sheet reflectivity, PMT reflectivity) in Lucid.
- Check Nahuel's notebook regarding the interpolation grid method for absorption correction.
- Show plots next week demonstrating how changing models affects the TOF distributions and event display.
Zhenxiong
- Prepare more explanation and diagrams on how the PMT angular response is calculated and averaged, including the formulas and integration details.
- Plot the different PMT groups into different regions to check for consistent patterns across groups.
ryotaro
- Continue focusing on the LED bump gain study and integrate the data into Lucid.
Summary
PMT Angular Response Analysis Updates
Zhenxiong presented updates on PMT angular response analysis using WCTE cosmic data with the SIREN model. He explained the methodology of comparing pre-tuned and post-tuned SIREN responses by sampling source points on a uniform grid and computing visibility as a function of cosine theta. The analysis revealed that pre-tuned responses for different PMT groups were similar except for one group with green response, which Zhenxiong attributed to PMTs being located on the barrel versus caps due to tank curvature. While post-tuned responses showed similar tendencies to previous laser studies, significant differences remained, and Zhenxiong noted challenges in comparing the results with previous studies due to normalization differences.
Angular Response Normalization Discussion
The team discussed Zhenxiong's presentation on angular response normalization in WCTE studies, with Patrick expressing confusion about the different normalization methods and asking for clarification on the calculations. Zhenxiong explained that he was averaging visibility rather than summing it, and described his method of averaging by PMT group for each cosine theta bin. The group identified discrepancies between Zhenxiong's results and previous Cummings studies, with Patrick suggesting they need to carefully consider how their optical siren analysis compares to traditional methods. Zhenxiong agreed to prepare more detailed explanations of his methodology and normalization approach.
Physics Models Implementation Progress Update
Riya presented updates on implementing physics models in Lucid to align with SK's internal models, including PMT and black sheet reflectivity. She showed time-of-flight comparison plots between Lucid simulation and SKDetsim, noting improvements but still observing differences that may be due to absorption, PMT geometry, and electronics responses. Riya also demonstrated initial fitting of optics parameters using a new loss function, with mixed results on constraining Rayleigh scattering and Mie scattering lengths. The team discussed next steps including creating finer binning histograms, implementing Siren methods for absorption and PMT QE corrections, and conducting individual effect studies to better understand the discrepancies between Lucid and SKDetsim.
