Call for Papers

ProbNum26 will have a dedicated volume in the Proceedings of Machine Learning Research (PMLR), where accepted papers will be published. A submission should be up to 8 pages (or shorter) using dedicated style files (appendix excluded). The submission deadline is scheduled for the 5th of March 2026. Subject areas are methods, theory and applications of probabilistic numerics. The authors of accepted papers will present talks and/or posters at ProbNum 2026. Details are as follows.

Key Dates

The current schedule, subject to revision, for submission and the review process is as follows. All deadlines are anywhere on Earth.

• Submission Deadline: 5th March 2026
• Reviews Released: 16th April 2026
• Responses Due: 23th April 2026
• Decisions: 14th May 2026

Subject Areas

ProbNum 2026 welcomes submissions on methods, theory and applications in Probabilistic Numerics or broader fields involving probabilistic quantification of estimation errors of deterministic quantities (computational uncertainties). Examples of topics are as follows.

Methods (Algorithms)

• Probabilistic (Bayesian or non-Bayesian) numerical methods
  • such as Bayesian quadrature, Bayesian optimisation, probabilistic solvers of ODEs or PDEs, probabilistic numerical linear algebra, especially for Gaussian process regression.
• Black-box probabilistic numerics.
• Reproducing kernel-based methods for problems of numerical analysis (corresponding to probabilistic numerical methods)
  • such as interpolation, quadrature, maximum mean discrepancy, global optimisation, differential equation solvers, physics-informed learning.

Theory

• Error bounds and convergence rates of probabilistic numerical methods.
• Well-calibratedness of uncertainty estimates.
• Properties of hypothesis spaces (e.g., Gaussian processes, reproducing kernel Hilbert spaces) of numerical methods.

Applications

• Computation-aware uncertainty quantification for
  • simulation, including the solution of (partial, ordinary, stochastic, and differential algebraic) differential equations for time-evolving processes;
  • deep learning-based simulation methods, superresolution methods, diffusion models, neural operators, and other algorithms aiming to functionally replace numerical computation.
• Computation-aware optimal planning and decision-making under uncertainty, including for control, robotics, active learning, and so on.
• Inverse problems and data assimilation, for example in scientific simulation.
• Applications in science and engineering, including (but not limited to) material science, physics and astronomy, climate science, geoscience, and finance.

Submission Instructions

Page limit
Submissions are full papers limited to up 8 pages excluding references, acknowledgements, and appendices. Shorter submissions are very welcome and will be equally considered. Any appendices may be submitted as part of the main pdf or separately as supplementary material.

Formatting instructions
Submissions must use the ProbNum LaTeX style package. Detailed instructions can be found in the file FormattingInstructions.tex in this zip file. Please do not modify the style file. Formatting instructions are available in the sample paper provided with the style package.

Anonymization
The ProbNum26 review process is double-blind. All submissions must be anonymized and may not contain any information that can violate the double-blind reviewing policy, such as the author names or their affiliations, acknowledgements, or links that can infer any author’s identity or institution. Self-citations are allowed as long as anonymity is preserved.

Submission page
The submission will be done via OpenReview. The OpenReview site will open by 30th January 2026. Please upload a single file; you can either submit a single pdf file or a single zip file for further supplementary material in other formats.

Dual submissions
Submitted manuscripts should not have been previously published in a journal or in the proceedings of a conference, and should not be under consideration for publication at another conference at any point during the ProbNum 2026 review process. This excludes non-archival venues such as workshops.

Confidentiality
Reviewer will be instructed to keep them confidential during the review process and delete them once the review process has concluded.

Reviewer Nomination

For each submission, the authors will be expected to nominate at least one of the authors as a reviewer for ProbNum26. Nominated reviewers are expected to have sufficient expertise in the relevant field.

Past Proceedings

• ProbNum 2025: PMLR 271