denehoffman / cow-factors

Implementation of various statistical weighting methods for particle physics analysis

About

This Rust program contains a simple analysis and comparison of multiple statistical weighting methods, including inPlot/sPlot¹ and Q-Factors². It is intended to show the need for a correction in the Q-Factors method, which occasionally fails because it does not use the proper event weighting scheme.

Usage

1. Clone the repository: git clone [e-mail]:denehoffman/cow-factors.git
2. Install the executable: cargo install --path cow-factors
3. Run the analysis script: cow-factors run output.tsv
4. Check the results: cow-factors process output.tsv results.tsv
5. Profit?

SLURM Usage

[WIP]

Plotting

[WIP]

TODO

• Write SLURM script and plotting scripts (Python)
• Implemement COW³ weights

Disclaimer

This analysis is part of an ongoing project which will eventually be published, please don't scoop me, I'll know it and raise hell :)

References

Footnotes

1. M. Pivk and F.R. Le Diberder. “sPlot: A statistical tool to unfold data distributions”. In: Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 555 (1-2 Dec. 2005), pp. 356–369. issn: 01689002. doi: 10.1016/j.nima.2005.08.106. arXiv:physics/0402083 ↩

2. M Williams, M Bellis, and C A Meyer. “Multivariate side-band subtraction using probabilistic event weights”. In: Journal of Instrumentation 4 (10 Oct. 2009), P10003–P10003. issn: 1748-0221. doi: 10.1088/1748-0221/4/10/P10003. arXiv:0809.2548 ↩

3. H Dembinski, M Kenzie, C Langenbruch, and M Schmelling. "Custom Orthogonal Weight functions (COWs) for Event Classification". In: Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment Volume 1040 (2022) 167270, issn: 0168-9002. doi: 10.1016/j.nima.2022.167270. arXiv:2112.04574 ↩