Probabilistic learning on manifolds constrained by nonlinear partial differential equations for small datasets
This work addresses the challenge of handling small datasets in stochastic computational models for engineers and scientists dealing with complex PDEs, though it appears incremental as an extension of existing PLoM methods.
The paper tackles the problem of synthesizing solutions to nonlinear stochastic PDEs with limited training data by extending Probabilistic Learning on Manifolds (PLoM) to incorporate PDE constraints, resulting in validated applications across uncertain dynamical systems, 2D Navier-Stokes flows, and 3D elastic structures.
A novel extension of the Probabilistic Learning on Manifolds (PLoM) is presented. It makes it possible to synthesize solutions to a wide range of nonlinear stochastic boundary value problems described by partial differential equations (PDEs) for which a stochastic computational model (SCM) is available and depends on a vector-valued random control parameter. The cost of a single numerical evaluation of this SCM is assumed to be such that only a limited number of points can be computed for constructing the training dataset (small data). Each point of the training dataset is made up realizations from a vector-valued stochastic process (the stochastic solution) and the associated random control parameter on which it depends. The presented PLoM constrained by PDE allows for generating a large number of learned realizations of the stochastic process and its corresponding random control parameter. These learned realizations are generated so as to minimize the vector-valued random residual of the PDE in the mean-square sense. Appropriate novel methods are developed to solve this challenging problem. Three applications are presented. The first one is a simple uncertain nonlinear dynamical system with a nonstationary stochastic excitation. The second one concerns the 2D nonlinear unsteady Navier-Stokes equations for incompressible flows in which the Reynolds number is the random control parameter. The last one deals with the nonlinear dynamics of a 3D elastic structure with uncertainties. The results obtained make it possible to validate the PLoM constrained by stochastic PDE but also provide further validation of the PLoM without constraint.