Request for contributions: PDESystemLibrary.jl - A library of systems of symbolic partial differential equations

Announcing PDESystemLibrary.jl, a library of symbolically defined PDESystems using ModelingToolkit.jl.

We have a few (over 400) systems already included as a start, with more being added all the time. We really want to include more systems that people are solving in practical research, so if you have a well posed example please consider reaching out in an issue, or submit a PR.

This library will be used across the SciML ecosystem in automated benchmarking of PDE discretizers such as NeuralPDE and MethodOfLines, leading to development in areas where we find they fall short. So if you want to know numerically what is the fastest/most accurate way to solve your favourite system of PDEs and ones like it, and drive the development of new and better solving methods, please get involved!


this is a very interesting development. In the moment I have no time to contribute. An incentive to really try it out would be for me:

  • “PDE Systems gallery” - (e.g. documneter generated) list showing the definitions
  • a HOWTO describing how to adapt my (or anybody’s) PDE solver to serve as a backend

That sounds like a good idea, and is hopefully where we are going with this. We’re a little ways off having such a how to, but I hope that some parts of MethodOfLines will eventually be able to form part of a parser that will be able to get us to this point.