arXiv Machine Learning

FREESIA: Covariance-Aware Posterior Transport for Expressive and Scalable Data Assimilation

The paper introduces FREESIA, a training‑free, covariance‑aware posterior transport method for data assimilation that embeds forecast cross‑covariance into a flow‑based transport to recover unobserved states while preserving non‑Gaussian posterior structure. It combines an observation‑adaptive proposal with posterior correction, providing an asymptotically exact approximation of nonlinear posteriors and a Wasserstein error bound. Experiments on Double‑Well, Lorenz‑96, and Kolmogorov flow demonstrate that FREESIA captures complex posterior structures and achieves up to a 56% reduction in RMSE compared to the best baseline in sparse, nonlinear, non‑injective observation scenarios.

arXiv Machine Learning
Oct 5

A Unified Framework for Bayesian Data Assimilation with Generative Models and Observation Interpolants

A new framework for Bayesian data assimilation is presented that leverages pretrained stochastic interpolants, flow matching, and diffusion models as posterior samplers without retraining. By conditioning the interpolant path on observations, a shared likelihood‑score correction is applied to the drift or velocity, unifying stochastic and deterministic sampling methods. The approach yields exact posterior samples when the intermediate likelihood score is known, and practical approximations use a Gaussian surrogate with bias‑corrected mean and inflated covariance, enabling tractable high‑dimensional applications such as linear‑Gaussian dynamics, stochastic Navier‑Stokes, and urban airflow simulations with up to $O(10^4)$ degrees of freedom.

By Nikolaj T. M\"ucke, Benjamin Sanderse
arXiv Machine Learning
Sep 24

PR-Smoother: Simulator-Preserving Non-Gaussian Smoothing for Data Assimilation

PR‑Smoother is an amortized smoothing method that preserves the explicit use of a prescribed simulator in both the evidence lower bound and the variational family. It learns only future‑conditioned corrections to the simulator’s rollout, yielding a non‑Gaussian smoothing distribution that can jointly infer state, parameters, and sensor bias from observations alone. The approach recovers the exact smoother in deterministic and linear‑Gaussian limits and has been shown to capture multimodal posteriors in Lorenz‑96 and scale to 16,384‑dimensional Kolmogorov flow.

By Yuta Tarumi
arXiv Statistics ML
Sep 17

A Continuous-Time Ensemble Kalman-Bucy Smoother for Causal Inference and Model Discovery

The paper presents an ensemble Kalman–Bucy smoother (EnKBS) for continuous‑time data assimilation of nonlinear dynamical systems, reconstructing conditional distributions from ensemble moments without needing tangent‑linear or adjoint models. It demonstrates that EnKBS achieves exact smoothing mean and covariance in the infinite‑ensemble limit for linear‑Gaussian systems and incorporates regularization techniques like covariance localization and inflation for high‑dimensional problems. The method is applied to Bayesian inference of causal relationships in a dyadic trigger‑feedback model and to an iterative learning algorithm that uncovers the structure and hidden parameters of a reduced‑order model of midlatitude atmospheric circulation, all with small ensembles under partial observations.

By Zhang Jiang (University of Wisconsin-Madison), Marios Andreou (University of Wisconsin-Madison), Sebastian Reich (University of Potsdam), Nan Chen (University of Wisconsin-Madison)