Manifold-adapted radial basis functions for reduced-order modelling of chaotic flows

Abstract

Chaotic systems often evolve on a low-dimensional attractor whose geometry varies from one region to another. We propose a non-intrusive reduced-order model that reads this local geometry by clustering and uses it to shape a radial basis library whose kernels adapt to each region. Fitting the reduced velocity onto this library by one global regularised least-squares solve gives an explicit, differentiable vector field that reproduces the long-term statistics, that is, the invariant measure, without any use of the governing equations. Since a radial basis field decays away from the data and cannot by itself return an escaped state, the integration is stabilised by a kinematic corrector whose magnitude is reported as a measure of how far each result rests on the learned field rather than on the corrector. On Lorenz-63 the model recovers the attractor, its marginal densities, and the positive and neutral Lyapunov exponents, while under-recovering the strong transverse contraction. On Lorenz-96 its valid prediction time is competitive with tuned neural-network and reservoir-computing forecasters, and the invariant measure is reproduced on both the full state and a reduced observable. On the Kuramoto--Sivashinsky equation and the quasiperiodic Kolmogorov flow the model matches the energy distribution and spectrum of an intrusive quantised-local Galerkin model, and improves on a global Galerkin projection of the same dimension, without ever projecting the governing equations.