hep-statistics-serialization-standard

KO | EN

gitlite — search

#react #typescript #android #rust #deep-learning #linux #java #windows #kubernetes #vue #security #pytorch #mcp #open-source

hep-statistics-serialization-standard

(★ 13)

Definition of the HEP statistics serialization standard (HS3)

hep-statistics-serialization-standard 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# HS³ - HEP Statistics Serialization Standard [![Build & Deploy Documentation](https://github.com/hep-statistics-serialization-standard/hep-statistics-serialization-standard/actions/workflows/gh-pages.yml/badge.svg)](https://github.com/hep-statistics-serialization-standard/hep-statistics-serialization-standard/actions/workflows/gh-pages.yml)

Welcome to the High Energy Physics Statistics Serialization Standard.

**The project is in an early stage and currently resembles more of a
working group and not yet a standard.**

**Everything, including naming, folder structure etc is under
construction and open to discussion**

## Citation

We have a preprint paper on [arxiv](https://arxiv.org/abs/2606.01760).

If you use HS³ in your work please cite as:

```
@article{hs3preprint,
  title={HS$^3$: A Descriptive, Interoperable Serialization Standard for Statistical Models in High-Energy Physics},
  author={Burgard, Carsten and Schulz, Oliver and Stark, Giordon and Rembser, Jonas and Cello, Simon and Grunwald, Cornelius},
  year={2026},
  eprint = "2606.01760",
  archivePrefix = "arXiv",
}
```

# In short

The HEP Statistics Serialization Standard (HS³) defines standards of
different statistical procedures and results used in High Energy Physics
(HEP) in terms of human-and machine readable representations. Different
versions are defined with specifications and semantics that are
acknowledged by a committee.

View the current draft of the HS³ standard here: https://hep-statistics-serialization-standard.github.io/

Corresponding implementations to check the
validity of files are also provided, at a _best effort_ basis.
More information can be found in the corresponding projects.

# Why

There are two main motivation to have a code-independent representation
of the likelihood and inference results.

- It allows to publish the likelihood; a long-term goal in High Energy
  Physics experiments.

- A framework- and language independent representation allows to use
  different frameworks interchangeably; at least for reasonable
  complicated cases. It removes the dependency on code and reduces the
  need for maintenance of legacy projects.

# Goal

HS³ standardizes the machine- and human-readable serialization of all
components involved in model fitting as used in High Energy Physics.
This includes the definition of the model, the data as well as the loss
function. The aim is to provide on one hand a language and framework
agnostic, machine-reabable and on the other hand a human-readable,
publishable and preservable representation. This would allow to
interchange the fitting framework and decouple the reproducibility from
the frameworks lifetime.

The focus of this project is not on the long-term storage of large data,
which may be needed for an actual implementation, but rather to define a
common serialization format.

TODO: more description

# Standard versions

Until 1.x, the standard is considered unstable and may introduce
backwards incompatible changes.

After that, backwords compatibility is guaranteed to be maintained
inside the major releases (1.x, 2.x) and should be kept, if possible,
also between the major releases, but is not guaranteed in favor of a
cleaner standard.

# Participation

In order to submit ideas, proposals and examples, you can either start a
discussion using issues or add the document(s), if you have some, in the
proposals (draft or pending) folder and create a PR to discussion them.

If you are interest to become part of the core committee, please open an
issue. Anyone is allowed to join.

## Emoji-Reactions to Issues, PRs, Discussions and Comments

To keep thread notifications low and avoid comment spam, we use GitHub reactions as a structured tool for project governance. We invite everyone to use these specific emojis.

By standardizing our reactions, we can all communicate the status of a proposal, signal technical intent, or show appreciation instantly without cluttering the discussion thread:

*   ❤️ (`:heart:`): **Gratitude.** Thank you for your time and effort spent. This is greatly appreciated, but does *not* imply technical approval or disapproval of the content.
*   👀 (`:eyes:`): **Community Call to Action.** I have reviewed this and find it interesting. I am stepping back to invite more community members to share their opinions on this topic.
*   👍 (`:+1:`): **Approval.** This proposal has approval as-is.
*   👎 (`:-1:`): **Disapproval.** This requires significant changes, or a much stronger scientific/technical argument before it can be reconsidered.
*   🚀 (`:rocket:`): **Fast-Track.** This is a minimal, uncontroversial, or critical fix. Maintainers intend to merge this immediately skipping a formal vote.
*   🎉 (`:tada:`): **Consensus Reached.** Glad we got this settled and found a path forward, regardless of the original opposing viewpoints.
*   😄 (`:smile:`): **Uncontroversial / Positive Tone.** This signals that the proposal is straightforward, or appreciates the highly professional, constructive tone of the conversation.
*   😕 (`:confused:`): **Clarification Needed.** The technical proposal or argument is currently unclear. Please provide more precise language, data examples, or use cases so we can evaluate it.

# Contact

To stay updated, you can sign up for the e-group:
<hep-statistics-serialization-standard@cern.ch>

# Involved projects

## zfit

A model fitting library in pure Python. It\'s focus is on
customizability and strong model building.

[zfit on Github](https://github.com/zfit/zfit)

## pyhf

Pure-Python implementation of statistical model for multi-bin
histogram-based analysis and its interval estimation; HistFactory in
Python.

[pyhf on Github](https://github.com/scikit-hep/pyhf)

## RooFit

The Toolkit for Data Modeling with ROOT (RooFit) is a package that
allows for modeling probability distributions in a compact and abstract
way.

[RooFit](https://root.cern.ch/roofit)

## BAT.jl

The Bayesian Analysis Toolkit in Julia.

[BAT.jl on Github](https://github.com/bat/BAT.jl)

# Users

## ATLAS

The ATLAS Collaboration has published measurements in this format. They can be found by the HS3-Badge on [HEPdata](https://www.hepdata.net/search/?q=analysis%3AHS3&sort_by=latest&page=1&collaboration=ATLAS).

# Committee

The comittee is responsible for the acceptance and denial of new
proposals and has to approve a new version of the standard.

hep-statistics-serialization-standard

hep-statistics-serialization-standard 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

hep-statistics-serialization-standard

hep-statistics-serialization-standard 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README