Jacobian support

[saiyao233]

Thank you for creating this wonderful library.

I noticed that calculate_jacobian is not supported yet. Is there a timeline for when this feature will be available?

[balakumar-s]

This has come up in the past. Getting gradients for the Jacobian is slightly complicated. However, we can expose a non-differentiable Jacobian. Would that be useful?

[saiyao233]

Yes, that would be useful. Even a non-differentiable Jacobian can be valuable for many purposes, such as analyzing kinematics or dynamics for understanding system behavior

[AlanSunHR]

Is there any progress on exposing the non-differentiable Jacobian?

[jianglongye]

Any progress on this? A non-differentiable Jacobian would indeed be helpful for many use cases. Thank you!

[CRLqinliang]

@balakumar-s Can I utilize the autograd of PyTorch to get the gradient of the Jacobian? (If I could code the Jacobian formula, I might use the Screw theory to write the Jacobian.)