aiqbench analysis
This research applies simulation-based inference to calibrate subgrid parameters in ocean models, leveraging foundation models to efficiently quantify uncertainty where traditional methods fail. By compressing simulator outputs with PCA and comparing neural estimators against training-free tabular approaches, the study demonstrates a significant reduction in computational cost for inverse problems in climate modeling. The core advancement lies in using machine learning to solve ill-posed parameter estimation tasks without requiring tractable likelihood functions, directly advancing the intersection of AI and scientific simulation.
Calibrating subgrid parametrizations of single-column ocean models via simulation-based inference
arXiv:2609.13242v1 Announce Type: cross Abstract: Subgrid parametrizations of vertical mixing in ocean models depend on free coefficients that cannot be measured directly and must be calibrated against high-fidelity references such as large-eddy simulations (LES). Existing approaches return point estimates and leave the associated uncertainty unquantified, a limitation when the inverse problem is ill-posed or when distinct parameter configurations fit the data comparably well. Simulation-based inference (SBI) addresses exactly this: given a prior and access to the simulator, it approximates the full posterior over parameters without requiring a tractable likelihood, at a cost set by the number of simulator evaluations. We apply it to \texttt{tunax}, a JAX-based single-column ocean model, to calibrate the coefficients of its $k$--$\varepsilon$ closure. A blockwise PCA summary statistic compresses the simulator output along the depth axis while preserving its forcing--horizon--variable structure, making inference tractable at modest budgets. We compare neural posterior estimation and its sequential variants against a recent training-free approach built on a tabular foundation model. The latter recovers informative posteriors from a few hundred simulator calls, outperforming the trained estimators at every budget considered.
Abstract:Subgrid parametrizations of vertical mixing in ocean models depend on free coefficients that cannot be measured directly and must be calibrated against high-fidelity references such as large-eddy simulations (LES). Existing approaches return point estimates and leave the associated uncertainty unquantified, a limitation when the inverse problem is ill-posed or when distinct parameter configurations fit the data comparably well. Simulation-based inference (SBI) addresses exactly this: given a prior and access to the simulator, it approximates the full posterior over parameters without requiring a tractable likelihood, at a cost set by the number of simulator evaluations. We apply it to \texttt{tunax}, a JAX-based single-column ocean model, to calibrate the coefficients of its $k$--$\varepsilon$ closure. A blockwise PCA summary statistic compresses the simulator output along the depth axis while preserving its forcing--horizon--variable structure, making inference tractable at modest budgets. We compare neural posterior estimation and its sequential variants against a recent training-free approach built on a tabular foundation model. The latter recovers informative posteriors from a few hundred simulator calls, outperforming the trained estimators at every budget considered.
Atmospheric and Oceanic Physics (physics.ao-ph); Machine Learning (cs.LG)
arXiv:2609.13242 [physics.ao-ph]
(or arXiv:2609.13242v1 [physics.ao-ph] for this version)
https://doi.org/10.48550/arXiv.2609.13242 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 274 words
Related coverage
HyperODE: Zero-Shot Surrogate for Simulation and Inference of Dynamical Systems
arXiv · Aug 4, 2026
New method automatically finds errors in quantum states
Quantum Zeitgeist · Sep 7, 2026
News
Quantum Computing Report · Jun 7, 2026
Online learning of subgrid-scale models for quasi-geostrophic turbulence in planetary interiors
arXiv · Jul 29, 2026