neos

Upstream optimisation for downstream inference

https://github.com/gradhep/neos

Science Score: 85.0%

This score indicates how likely this project is to be science-related based on various indicators:

✓
CITATION.cff file
Found CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
✓
DOI references
Found 3 DOI reference(s) in README
✓
Academic publication links
Links to: arxiv.org, zenodo.org
✓
Committers with academic emails
1 of 9 committers (11.1%) from academic institutions
✓
Institutional organization owner
Organization gradhep has institutional domain (mattermost.web.cern.ch)
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (11.9%) to scientific vocabulary

Keywords from Contributors

jax asymptotic-formulas closember cls frequentist-statistics hep hep-ex high-energy-physics histfactory scikit-hep

Last synced: 6 months ago · JSON representation ·

Repository

Upstream optimisation for downstream inference

Basic Info

Host: GitHub
Owner: gradhep
License: bsd-3-clause
Language: Jupyter Notebook
Default Branch: main
Size: 64.7 MB

Statistics

Stars: 69
Watchers: 2
Forks: 8
Open Issues: 12
Releases: 11

Created about 6 years ago · Last pushed 7 months ago

Metadata Files

Readme Contributing License Citation

README.md

neos logo
neural end-to-end-optimised summary statistics
arxiv.org/abs/2203.05570

About

Leverages the shoulders of giants (jax and pyhf) to differentiate through a high-energy physics analysis workflow, including the construction of the frequentist profile likelihood.

If you're more of a video person, see this talk given by Nathan on the broader topic of differentiable programming in high-energy physics, which also covers neos.

You want to apply this to your analysis?

Some things need to happen first. Click here for more info -- I wrote them up!

Have questions?

Do you want to chat about neos? Join us in Mattermost:

Cite

Please cite our newly released paper:

@article{neos, Author = {Nathan Simpson and Lukas Heinrich}, Title = {neos: End-to-End-Optimised Summary Statistics for High Energy Physics}, Year = {2022}, Eprint = {arXiv:2203.05570}, doi = {10.48550/arXiv.2203.05570}, url = {https://doi.org/10.48550/arXiv.2203.05570} }

Example usage -- train a neural network to optimize an expected p-value

setup

In a python 3 environment, run the following: pip install --upgrade pip setuptools wheel pip install neos pip install git+http://github.com/scikit-hep/pyhf.git@make_difffable_model_ctor

With this, you should be able to run the demo notebook demo.ipynb on your pc :)

This workflow is as follows: - From a set of normal distributions with different means, we'll generate four blobs of (x,y) points, corresponding to a signal process, a nominal background process, and two variations of the background from varying the background distribution's mean up and down. - We'll then feed these points into the previously defined neural network for each blob, and construct a histogram of the output using kernel density estimation. The difference between the two background variations is used as a systematic uncertainty on the nominal background. - We can then leverage the magic of pyhf to construct an event-counting statistical model from the histogram yields. - Finally, we calculate the p-value of a test between the nominal signal and background-only hypotheses. This uses the familiar profile likelihood-based test statistic.

This counts as one forward pass of the workflow -- we then optimize the neural network by gradient descent, backpropagating through the whole analysis!

Thanks

A big thanks to the teams behind jax, fax, jaxopt and pyhf for their software and support.

Owner

Name: gradHEP
Login: gradhep
Kind: organization

Website: https://mattermost.web.cern.ch/signup_user_complete/?id=zf7w5rb1miy85xsfjqm68q9hwr
Repositories: 5
Profile: https://github.com/gradhep

Applying differentiable programming to high-energy physics. Join our mattermost chat with the link below, where we discuss + have irregular meetings!

Citation (CITATION.cff)

cff-version: 1.2.0
message: "Thanks for being interested in neos! If you use this software in a project, please cite it as below."
authors:
  - family-names: Simpson
    given-names: Nathan
    orcid: https://orcid.org/0000-0003-4188-829
  - family-names: Heinrich
    given-names: Lukas
    orcid: https://orcid.org/0000-0002-4048-7584
title: "neos: version 0.2.0"
version: v0.2.0
date-released: 2021-01-12
url: "https://github.com/gradhep/neos"
doi: 10.5281/zenodo.6351423
references:
  - type: article
    authors:
    - family-names: Simpson
      given-names: Nathan
      orcid: https://orcid.org/0000-0003-4188-829
    - family-names: "Heinrich"
      given-names: "Lukas"
      orcid: "https://orcid.org/0000-0002-4048-7584"
      affiliation: "TU Munich"
    title: "neos: End-to-End-Optimised Summary Statistics for High Energy Physics"
    doi: 10.48550/arXiv.2203.05570
    url: "https://doi.org/10.48550/arXiv.2203.05570"
    year: 2022

GitHub Events

Total

Watch event: 1
Push event: 18
Pull request event: 1
Fork event: 3

Last Year

Watch event: 1
Push event: 18
Pull request event: 1
Fork event: 3

Committers

Last synced: almost 3 years ago

All Time

Total Commits: 181
Total Committers: 9
Avg Commits per committer: 20.111
Development Distribution Score (DDS): 0.343

Top Committers

Name	Email	Commits
Nathan Simpson	e**n@g**m	119
Nathan Simpson	n**n@h**e	35
Matthew Feickert	m**t@c**h	5
pre-commit-ci[bot]	6**]@u**m	4
andrzejnovak	n**j@g**m	4
Lukas Heinrich	l**h@g**m	4
Nathan Simpson	p**e@p**m	4
dependabot[bot]	4**]@u**m	3
gehring	c**g@g**m	3

Committer Domains (Top 20 + Academic)

cern.ch: 1 hep.lu.se: 1

Issues and Pull Requests

Last synced: 7 months ago

All Time

Total issues: 10
Total pull requests: 27
Average time to close issues: about 1 month
Average time to close pull requests: 29 days
Total issue authors: 5
Total pull request authors: 6
Average comments per issue: 1.6
Average comments per pull request: 0.96
Merged pull requests: 21
Bot issues: 0
Bot pull requests: 11

Past Year

Issues: 0
Pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Issue authors: 0
Pull request authors: 0
Average comments per issue: 0
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

View more stats

Top Authors

Issue Authors

phinate (5)
andrzejnovak (2)
alexander-held (1)
lukasheinrich (1)
gehring (1)

Pull Request Authors

phinate (11)
dependabot[bot] (7)
pre-commit-ci[bot] (4)
matthewfeickert (2)
gehring (2)
andrzejnovak (1)

Top Labels

Issue Labels

enhancement (5) help wanted (3) dependencies (1) good first issue (1)

Pull Request Labels

dependencies (7) documentation (3)

Packages

Total packages: 1
Total downloads:
- pypi 66 last-month

Total dependent packages: 0
Total dependent repositories: 1
Total versions: 7
Total maintainers: 1

pypi.org: neos

UpUpstream optimization of a neural net summary statistic with respect to downstream inference goals.

Homepage: https://github.com/gradhep/neos
Documentation: https://neos.readthedocs.io/
License: BSD-3-Clause
Latest release: 0.3.0
published over 3 years ago

Versions: 7
Dependent Packages: 0
Dependent Repositories: 1
Downloads: 66 Last month

Rankings

Stargazers count: 8.4%

Dependent packages count: 10.1%

Forks count: 14.2%

Average: 16.4%

Dependent repos count: 21.6%

Downloads: 27.5%

Maintainers (1)

phinate

Last synced: 7 months ago

Dependencies

examples/requirements.txt pypi

celluloid *
plothelp *

.github/workflows/ci.yml actions

actions/checkout v1 composite
actions/upload-artifact v2 composite
pypa/gh-action-pypi-publish v1.4.2 composite

pyproject.toml pypi

setup.py pypi