Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

docs: Make 'Statement of Need' explicit section in JOSS paper #1290

Merged
merged 5 commits into from
Feb 4, 2021
Merged

docs: Make 'Statement of Need' explicit section in JOSS paper #1290

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
c1743b1
Add 'Statement of Need' as a section
matthewfeickert Feb 4, 2021
61c29d7
Move source code archival and version ouf of Statement of Need
matthewfeickert Feb 4, 2021
558ce24
Merge branch 'master' into docs/make-statement-of-need-a-section
kratsg Feb 4, 2021
4f2cbb2
Add comma
matthewfeickert Feb 4, 2021
8840586
Merge branch 'master' into docs/make-statement-of-need-a-section
kratsg Feb 4, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions docs/JOSS/paper.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,13 +40,15 @@ The relationship between them is often formalised in a statistical model $f(\mat
Given observed data, the likelihood $\mathcal{L}(\mathbf{\phi})$ then serves as the basis for inference on the parameters $\mathbf{\phi}$.
For measurements based on binned data (histograms), the `HistFactory` family of statistical models [@Cranmer:1456844] has been widely used in both Standard Model measurements [@HIGG-2013-02] as well as searches for new physics [@ATLAS-CONF-2018-041].
`pyhf` is a pure-Python implementation of the `HistFactory` model specification and implements a declarative, plain-text format for describing `HistFactory`-based likelihoods that is targeted for reinterpretation and long-term preservation in analysis data repositories such as HEPData [@Maguire:2017ypu].
The source code for `pyhf` has been archived on Zenodo with the linked DOI: [@pyhf_zenodo].
At the time of writing this paper, the most recent release of `pyhf` is [`v0.5.4`](https://doi.org/10.5281/zenodo.4318533).

# Statement of Need

Through adoption of open source "tensor" computational Python libraries, `pyhf` decreases the abstractions between a physicist performing an analysis and the statistical modeling without sacrificing computational speed.
By taking advantage of tensor calculations, `pyhf` outperforms the traditional `C++` implementation of `HistFactory` on data from real LHC analyses.
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would start with the need for a pure python version that is independent of ROOT. Then mention it's faster. And say that yaml aids in likelihood publishing. Not totally sure what "decreases the abstractions between ..." means here.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

well, I wrote this a while ago, but didn't click "submit review". Paper is published! So ignore.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I think we got stuck between vague requirements and having the JOSS take a while in this case. It is a fair balance to just repurpose existing text -- but your point is noted. I doubt we can make that phrasing change.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry this didn't get posted in time, but thanks for still adding this comment though! Still good to have for the future (as more JOSS papers will come from IRIS-HEP and Scikit-HEP).

`pyhf`'s default computational backend is built from NumPy and SciPy, and supports TensorFlow, PyTorch, and JAX as alternative backend choices.
These alternative backends support hardware acceleration on GPUs, and in the case of JAX JIT compilation, as well as auto-differentiation allowing for calculating the full gradient of the likelihood function — all contributing to speeding up fits.
The source code for `pyhf` has been archived on Zenodo with the linked DOI: [@pyhf_zenodo].
At the time of writing this paper the most recent release of `pyhf` is [`v0.5.4`](https://doi.org/10.5281/zenodo.4318533).

## Impact on Physics

Expand Down