GPDiffEq.jl

The GPDiffEq.jl package facilitates learning the (nonlinear) vector field of unknown system using Gaussian Processes (GPs).

It bridges the Universal Differential Equations in the SciML community using Neural Networks and GP ecosystem by the JuliaGaussianProcesses organization.

For the GP machinery, this package builds on

• KernelFunctions.jl
• ApproximateGPs.jl
• InducingPoints.jl.

It further uses Flux.jl for training, and DifferentialEquations.jl for solvers.