🚀 The feature, motivation and pitch

Hi!

Recently I am working on a project called torchcontrol, which aims implementing parallel control system simulation and control library based on PyTorch.

At the begining, I also tried to use state transition to calculate the new state. But I found it usually needs to discretelize the continuous time system, which is not easy to be done parallelly. So I turn to use odeint from torchdiffeq. This method provides a more accurate solution and is also differentiable.

I've also done the code here torchcontrol/plants/PlantBase.step. You can click and check.

Thanks!

Alternatives

Hi there!

I am wondering whether this repo still accept contribution to existing codes, such as module.System.

I would like to merge my code in torchcontrol into this repo if you don't mind it.

Thanks!

Additional context

No response