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 reporting progress on various simulation and calibration projects. Zhenxiong presented updates on cosmic calibration with WCTE real data, including code updates to the repository and analysis of PMT reconstruction using lookup tables. She discovered that PMT saturation was set to 30 and found this was causing bias in the observed charge comparisons between pre-tuned and post-tuned models. SLAC suggested checking how saturation is implemented and recommended analyzing 1D charge distributions to better understand the effect. Riya then shared progress on recreating Super-K's time-of-flight plots using Lucid simulation, comparing their results to published data and discussing plans to create SK4 Monte Carlo simulations for calibration. Finally, SLAC provided feedback on the structure and timeline for a simulation paper being developed, suggesting to aim for completion by end of July to meet an August deadline.
Next steps
Patrick/Zhenxiong
- Check with WCTE experts (e.g., Claude) on how PMT saturation is treated in MDT/WCSim.
Riya
- Document the physics models in Lucid and compare/align them with Super-K's internal models.
- Produce SK4 Monte Carlo simulation using SKDetsim for calibration fits.
- Compare basic distributions and Lucid fit results to MC input parameters.
Zhenxiong
- Double check and reset the PMT saturation value in the cosmic calibration pipeline, comparing with WCTE class and checking the effect on the charge scale plots.
- Process the beam data and run the reconstruction after fixing the PMT saturation issue.
- Write documentation on how to use the scripts in the repository.
Junjie
- Junjie/Sam: Aim to complete the detector simulation paper by end of August, starting with a brain dump of content.
Summary
Cosmic Calibration Updates Presentation
Zhenxiong presented updates on cosmic calibration work with WCTE real data, including code updates in the repository and a pull request for the cosmic collaboration branch. She reported on the pre-tune reconstruction using lookup tables for Cherenkov profiles up to 10 GeV, finding that using lookup tables did not improve the readout compared to the previous SIREN model, inconclusive for explaining the previous high-charge bias. Zhenxiong also shared results from cosmic calibration with real data using 1 GeV Cherenkov profiles and cosmic data.
SIREN Training and PMT Saturation
Zhenxiong discussed issues with SIREN training and charge bias, explaining that the lookup table used coarse profiles and initial momentum values affected the results. The team examined PMT saturation effects, with Zhenxiong noting that saturation was applied during the tuning process but not in the pre-tuned simulation, which could explain why saturation effects appeared only in the post-tuned results. SLAC suggested analyzing the color 3D distribution rather than relying on 1D spread plots, and recommended checking the PMT saturation value against WCTE settings. Zhenxiong also presented reconstruction results showing that pre-tuned results were closer to fiTQun results than post-tuned results, though she was uncertain about the cause.
PMT Saturation Issues Discussion
The team discussed issues with PMT saturation in their data analysis, where Zhenxiong identified a problem with PMT saturation being set to 30 in the configuration file. SLAC suggested checking for saturation effects by examining 1D plots to look for dips in the data distribution, and recommended increasing the saturation value if no saturation effects are observed. The team also discussed the need to consult with WCTE experts about appropriate saturation settings, with Patrick planning to check with Claude for insights on how saturation is handled in MDT modeling.
Time-Of-Flight Plot Simulation Analysis
Riya presented her work on recreating a time-of-flight plot from Super-K calibration data using Lucid simulation, comparing the results to published data. She identified discrepancies between the simulated and actual plots. Riya explained she is investigating potential causes including reflection models, PMT timing response, and index of refraction, while also working to document and align Lucid physics models with SK's internal models. SLAC suggested exploring whether Riya could optimize parameter values directly from the data points rather than using the same values as the SK model, which Riya confirmed was part of her planned approach using SK4 Monte Carlo simulation.
Photon Detection Model Validation
The team discussed data fitting and model validation approaches for photon detection simulations. SLAC suggested focusing on simpler data subsets, particularly single PMT timing distributions from known light sources, to validate models before working with more complex integrated data. The group also reviewed progress on a simulation paper, which will cover photon propagation, electronics, and performance applications including an Open Dataset for DORAEMON. The team agreed to aim for completing the paper by end of July to meet an August deadline.
