TorchDAE: Implicit DAE Solvers with Index Reduction and Adjoint Sensitivity

Hello everyone,

I have been working on a library for solving Differential Algebraic Equations in PyTorch since there haven’t been any solvers with vectorization or GPU-accelerated computations for them.

The library includes algorithms that aren’t implemented in any Python ecosystem, including Generalized Alpha, Dummy Derivatives, and adjoint sensitivity for DAEs.

Would love to hear some feedback from the PyTorch Community.

Github Repo: GitHub - yousef-rafat/torchdae: Differential-Algebraic Equation solver in PyTorch. Differentiable and GPU-enabled · GitHub

looks absolutely amazing! I am trying to build a Modelica inspired DAE assembler in Python with Pytorch as one of the target backends. I would love to chat! GitHub - AN-Best/braid · GitHub