IRIS-HEP Fellow: Daryna Shevchuk
Fellowship dates: Jun – Aug, 2026
Home Institution: Ukrainian Catholic University
Project: Benchmarking and Optimizing PyHS3
Develop a comprehensive benchmarking and optimization infrastructure for PyHS3, a pure-Python implementation of the HEP Statistics Serialization Standard (HS3). The project focuses on measuring runtime performance and memory usage across representative High Energy Physics statistical workloads, including probability density function evaluation, likelihood calculation, parameter fitting, and model scaling. Benchmarking results will be compared against established frameworks such as ROOT, pyhf, zfit, and numba-stats to ensure both performance improvements and numerical agreement. Using profiling-driven analysis, the project will identify bottlenecks in graph construction, compilation, and model evaluation, and implement optimizations such as caching and computational graph simplifications. The resulting infrastructure will provide reproducible benchmarks, continuous performance tracking, and documentation to support future PyHS3 development.
More information: My project proposal
Mentors:
-
Giordon Stark (University of California, Santa Cruz)
Current Status
Contact me: