Estimating Space-Dependent Coefficients for 1D Transport Using Gaussian Processes as State Estimator in the Frequency Domain

This letter presents a method to estimate the space-dependent transport coefficients (diffusion, convection, reaction, and source/sink) for generic scalar model, e.g., heat or mass. As problem is solved in frequency domain, complex valued state as function of spatial variable estimated using Gaussian process regression. The resulting probability density state, together with semi-discretization linear parameterization are used determine maximum likelihood solution these coefficients. proposed illustrated by simulations.