Random-Feature Kalman Filtering for Linear PDE Data Assimilation

Abstract

Data assimilation for time-dependent partial differential equations (PDEs) requires Bayesian updates of an evolving field from streaming, sparse, and noisy observations, while keeping the filtering state finite dimensional. We introduce a random-feature Kalman filtering framework for linear PDE data assimilation. Once the random features are frozen and the linear PDE is Galerkin discretized, the coefficient vector satisfies a finite-dimensional linear-Gaussian state-space model, so the Kalman recursion gives the exact posterior for the chosen coefficient model. For non-orthogonal random-feature draws, we construct a mass-whitened effective-rank coordinate system that removes near-null mass directions and identifies the posterior dimension $r$. For the heat equation with implicit-Euler time stepping, we prove a high-probability posterior-contraction and PDE-consistency theorem in these mass-whitened coordinates. The mean-square $L^2$ reconstruction error separates into an effective-rank feature approximation term, a deterministic time-consistency term, and a Bayesian estimation term. In the high-information regime, the leading posterior contribution scales as $rσ^2/N_o$, where $σ^2$ is the observation-noise variance and $N_o$ is the number of observations per analysis time. Thus the analysis distinguishes the exact coefficient-space posterior from deterministic PDE approximation errors, and gives a checkable uncertainty-quantification guarantee for random-feature filtering of a representative parabolic PDE.