Submission instructions and review process

General guidelines:

Our workshop is designed to highlight exciting progress and interesting emerging research directions at the intersection of machine learning and the physical sciences for the benefit of researchers in our community. In an effort to reflect the strengths and preferences of our community as a whole, we rely on this community – i.e. humans with experience in the field – to help us select papers based on their personal judgments of correctness, novelty, significance, and potential impact. Reviewing is a critical yet relatively thankless act of service, and we have structured our call for papers out of respect for our reviewers’ time and efforts.

  • Papers should represent rigorous human effort and decision-making in both the analysis and the reporting. Obvious artifacts of unverified generative AI use can undermine a reviewer’s confidence in the rigor of a submission. To aid reviewers in their assessments of scientific rigor, authors who used generative AI at any point in their process are strongly encouraged to not only describe how AI was used in the preparation of a work, but also how they verified the accuracy of any generated outputs.

  • Papers should demonstrate high standards of accuracy and reproducibility. Authors are responsible for ensuring that the content of their paper (including all text, diagrams, figures, code, and citations) thoroughly and accurately describes their methods and results so that the reviewers can assess the technical validity of the work as a whole. If authors choose to use AI models to generate content such as text or images, they do so at their own risk, as these tools can introduce inaccuracies that may be grounds for paper rejection. Papers containing erroneous citations – e.g. discrepancies between cited and linked paper titles, author lists, etc. – will not be sent out for review.

  • Papers should be concise. Submissions should respect reviewers’ time by not exceeding 4 pages in length. Appendices are discouraged, and reviewers have no obligation to read them when submitting reviews.

  • Papers should be enjoyable to read. Papers should be engagingly and clearly written, using legible and thoughtfully-constructed figures as appropriate, to convince reviewers of the work’s significance and provide deeper insights where possible. Papers should be approachable for a reviewer who is not an expert in your specific area of physical science, and should therefore avoid or at least define any technical jargon.


    • Specific guidelines:

  • Submitted papers must use LaTeX formatting using the NeurIPS 2026 template. Papers submitted without using this template will be desk rejected.
  • Authors should replace the footer text in the LaTeX template with the following: “Submitted to the 9th Workshop on Machine Learning and the Physical Sciences (ML4PS 2026). Do not distribute.” Any other modifications to the LaTeX template will result in a desk rejection.
  • The NeurIPS paper checklist is not required for submissions to this workshop.
  • The review process is double-blind (optionally, single-blind for the Evaluations & Benchmarks track). Outside of the Evaluations & Benchmarks track, papers must be fully anonymized. This means that code, links, and all text and figures should be anonymized as well. Authors are welcome to use tools such as Anonymous Github to share anonymized code with reviewers.
  • All authors must be registered in the submission system at the time of submission. We will not allow authors to be added after the review process has begun.
  • This workshop is not archival and there are no published proceedings. While we primarily encourage the submission of original works, we also accept submissions that are extended abstract versions of published work if the topic fits particularly well with the workshop's scope. However, such papers will likely need to be rewritten to fit our format and venue.
  • Incomplete works at an advanced progress stage are welcome, as are negative or null results that add value and insight.
  • We will award a Reproducibility Prize to a small number of submissions containing exceptionally well-documented code and reproducible workflows.
  • Papers should be submitted via the workshop's OpenReview page.


    • Grounds for desk rejection:

  • Papers that do not use the NeurIPS 2026 LaTeX template.
  • Papers that modify the NeurIPS 2026 LaTeX template in any way (including but not limited to adjusting margin widths, font sizes, etc.), other than replacing the footer text.
  • Papers (outside of the Evaluation & Benchmarks track) that are not anonymized (e.g. author names are listed or identifiable information such as GitHub usernames are included). Note that all code, links, text, and figures should be fully anonymized.
  • Papers that are longer than 4 pages (not including references).
  • Papers containing any erroneous citations (e.g. discrepancies between cited & linked titles, author lists, etc.)


    • Review process

    Submissions that follow the submission instructions correctly (i.e., are not rejected due to editorial reasons, such as exceeding the page limit or tampering with the template format) are sent for peer review. The review process takes place on OpenReview. Below are some of the key points about this process that are shared with the reviewers and authors alike. Authors are expected to consider these in preparation of their submissions and when reviewing.

    • There will be multiple reviewers for each paper.
    • Reviewers will be able to state their confidence in their review.
    • We will provide an easy-to-follow template for reviews so that both the strengths and weaknesses of the submission can be highlighted.
    • Reviewers will be matched with papers via OpenReview, which handles potential conflicts-of-interest based on home institution and author collaborations.
    • Criteria for a successful submission include novelty, correctness, relevance, and potential impact.
    • Minor flaws will not be the sole reason to reject a paper.
    • There will be no rebuttal period.