USCMS Researcher: Lauren Hay
Postdoc dates: Apr 2026 - Mar 2027
Home Institution: Brown University
Project: Reducing storage needs through negative weight mitigation in MC events and lossy compression
Aim to reduce tape and disk storage needs through developing slimmer data formats by removing or compressing redundant/and or unused information. Additionally reduce disk storage needs by improving efficiency of MC generators by mitigating negative weights after generation.More information: My project proposal
Mentors:
-
Matt LeBlanc (Brown University)
- 26 May 2026 - "Cell Reweighting Algorithms for Pathological Weight Mitigation in LHC Simulations using Optimal Transport", Lauren Hay, CHEP 2026
- 17 Jul 2026 - "Resampling Away Negative Weights with Optimal Transport", Lauren Hay, CHEP 2026
Current Status
2026 Q2
- Progress
- Finalized studies of OT based cell-resampling of negative weights in generic herwig samples.
- Presented results at CHEP 2026.
- First attempt at training AE for compressing particle candidates.
- Tried MSE and sliced-Wasserstein based loss functions with some success.
- Still have issues in deconding constituent pT’s – this causes an underestimation of jet mass and pT.
- Likely need to try an alternative architecture
- Finalized studies of OT based cell-resampling of negative weights in generic herwig samples.
- Began investigating lossy compression implementation in CMSSW and it’s physics performance.
Contact me: