About GSoC

[DMax1314]

Hi, this is Zhendong Li form Shanghai University. I am wringing to ask questions about GSoC project Benchmarks and Tutorial Writing.

1. Could you please give more details about how to take part in this project?
2. What can I do right now to know or contribute to this project?
3. Do I need to try any tests and give solutions of tests? Many thanks!

[ChrisRackauckas]

Hey, are you at SIAM CSE right now? If so we could chat.

Could you please give more details about how to take part in this project? What can I do right now to know or contribute to this project?

Most of the tutorials have been moved to the respective libraries by now. So things like the physical discovery ones are now in https://docs.sciml.ai/Overview/stable/showcase/missing_physics/ .

I think a good project in this area is to update the examples from https://arxiv.org/abs/2001.04385 . The very old code lives in https://github.com/ChrisRackauckas/universal_differential_equations but we want to make those all documentation examples for SciMLSensitivity.jl. So a folder in https://github.com/SciML/SciMLSensitivity.jl/tree/master/docs/src/examples with UDE examples would be fantastic. The missing physics example is one of those, so you can ignore that one, but the PDE discovery etc. is all relevant. Along with the examples from https://arxiv.org/abs/2012.07244 where the files currently live in https://github.com/RajDandekar/MSML21_BayesianNODE. Getting those all into standard docs examples would be fantastic, and a good way to learn the SciML automated discovery tooling.

Do I need to try any tests and give solutions of tests?

I think the best "test" is really just to get started. I would start with the Fisher-KPP example and try to get an updated working version of that which follows what you see in the missing physics one. So set it up with Lux.jl, get everything training again, etc. Ask if you have any questions in the Zulip or Slack.