pyhf: a pure-Python statistical fitting library with tensors and automatic differentiation
Nov 26, 2022
6 pages
Published in:
- PoS ICHEP2022 245
Contribution to:
- Published: Nov 26, 2022
e-Print:
- 2211.15838 [hep-ex]
DOI:
View in:
Citations per year
Abstract: (SISSA)
The HistFactory p.d.f. template is per-se independent of its implementation in ROOT and it is use- ful to be able to run statistical analysis outside of the ROOT, RooFit, RooStats framework. pyhf is a pure-Python implementation of that statistical model for multi-bin histogram-based analy- sis and its interval estimation is based on the asymptotic formulas of “Asymptotic formulae for likelihood-based tests of new physics”. pyhf supports modern computational graph libraries such as TensorFlow, PyTorch, and JAX in order to make use of features such as auto-differentiation and GPU acceleration. In addition, pyhf’s JSON serialization specification for HistFactory models has been used to publish 23 full probability models from published ATLAS collaboration analyses to HEPData.Note:
- 6 pages, 1 figure, 1 listing. Contribution to the Proceedings of the 41st International Conference on High Energy physics (ICHEP 2022). If you are looking to cite pyhf as software, please follow the citation instructions at https://pyhf.readthedocs.io/en/stable/citations.html
- model: statistical
- acceleration
- ATLAS
- new physics
- statistical analysis
- data analysis method
- programming
References(18)
Figures(1)
- [1]
- [2]
- [3]
- [4]
- [5]
- [6]
- [7]
- [8]
- [9]
- [10]
- [11]
- [12]
- [13]
- [14]
- [15]
- [16]
- [17]
- [18]