This repository contains the code for the paper Learning ODE Models with Qualitative Structure Using Gaussian Processes, CDC 2021 paper WeA01.3, presented on Wednesday 15.12.21 at 13:30 UTC.
The packages in this repository will very soon be superseded by GPDiffEq.jl.
This code is compatible with Julia 1.5 To recreate the environment open a terminal in the repository and run
julia --projector within julia
] activate .Then run
] instantiate
to install and build all packages in the environment.
The paper covers two examples, a simple spiral ODE and the Kepler problem.
The plots below are produced by the plots_spiral.jl script, and written out into the plots directory. The plots are generated via precomputed data from the data directory.
The data can be regenerated by running the setup_spiral.jl script. Note that the computations are set up to be multi-threaded, and took about 30min on 12 threads. In addition, they require substantial RAM, at about 18Gb when using 12 threads.
The plot below is produced by the plots_kepler.jl script, and written out into the plots directory. The plot is generated via precomputed data from the data directory.
The data can be regenerated by running the setup_kepler.jl script.
