search

SciML / diffeqpy

Solving differential equations in Python using DifferentialEquations.jl and the SciML Scientific Machine Learning organization
MIT License
531 stars 39 forks source link

Feature request: Ingest sympy representation of ModelingToolkit and return PyTorch tensor with gradients and events #115

Open djinnome opened 1 year ago

djinnome commented 1 year ago

@djinnome says:

If we can represent the ODE symbolically as something that the Modeling toolkit can handle, then symbolic gradients and events computed in Julia can be passed back to PyTorch?

Should we create an issue on DiffEqPy?

On August 16, 2023, @ChrisRackauckas said

Sure. I think there’s a few issues asking for exactly this same thing already though. Whenever I go to a conference I hear this question.

@djinnome says: Seems like issues #57 and #67 are fairly old open issues that seem to touch upon the input and output aspects of this issue.

@ChrisRackauckas wrote:

modelingtoolkitize(prob)
ODEProblem(sys, dstar_jac=true)

@djinnome says: Can you add a github action to expose this issue on the DARPA-ASKEM integration project by tagging this issue with the integration label?