taichi VS dolfinx

The haughtiness is not for nothing. Since Dec 2023, they made a lame excuse that Pytorch didn't support 3.12: https://github.com/taichi-dev/taichi/issues/8365#issuecommen...

Later, even when Pytorch added support for 3.12, nothing changed (so far) in Taichi.

taichi – Productive, portable, and performant GPU programming in Python

ETH Zurich is using it for their physics sim courses, University of Utah is using it for simulations (SIGGRAPH 2022), OPPO (they make smart devices running Android), Kuaishou uses it for liquid and gas simulation on GPUs. Lots of GPU accelerated sim stuff.

https://www.taichi-lang.org/

https://www.researchgate.net/publication/337118128_Taichi_a_...

https://github.com/taichi-dev/taichi

Check our the realease note (https://github.com/taichi-dev/taichi/releases) for more improvements.

I've recently started using Taichi (https://taichi-lang.org/) for numerical codes and the fact it doesn't try to trick you into thinking it's numpy is a nice "feature". ;)

Taichi v1.4.0 is released! See what's new: - Taichi AOT, along with a native Taichi Runtime library: Native applications can now load compiled AOT modules and launch Taichi kernels without a Python interpreter. - Taichi ndarray: An array object that holds contiguous multi-dimensional data to allow easy data exchange with external libraries. - Dynamic index: Use variable indices whenever necessary on all backends without affecting the performance of those matrices with only constant indices. See deprecation and more improvements in the release note.

One thing that is a pain when learning this stuff is that actually performing the optimization requires a good understanding of the numerical discretization of PDEs. Finite elements are a natural choice because it is very easy to characterize the adjoint with this formulation. There are some good free tools that you can use to actually learn and do some computations yourself. The first is hIPPYlib (paper, code), which is built on top of FEniCS (link), for which there are many good tutorials. Beware trying to install this on Windows though. You will need to work in Docker or in Ubuntu via Windows Linux Subsystem.

If you're ok with coding, fenics is a solid place to start. Also if you're comfortable with coding, openfoam is FVM, rather than FEM, but it can handle solidmechanics.

> FEniCs made FEM so easy

https://fenicsproject.org/

Indeed, was blown away when I saw it for the first time over a decade ago, compared to the convoluted C++ FEM libraries I had seen before that.

Have you looked at FEniCS? Pretty much everything else I'm aware of is probably overkill (e.g., MOOSE in C++, HYPRE's Python bindings, etc.)

FEniCSx is quite good.

You mean Python? For many research tasks it's fine. High level libraries let you define your computation in a minimal amount of code. FEniCS is a great example of this - underneath it compiles the abstracted high level stuff to calls to low-level libraries that do the heavy lifting. For many applications you can just write vectorized code with Numpy that performs well, or use Numba to JIT what you can't vectorize. For some tasks, however, you need interfaces that don't exist in the high level libraries, and that was the case for me.

I just came across FEniCSX. I’m not sure if it’s what you want but here’s the description:

Fenics: Very popular finite element framework “UseTab: Never” https://github.com/FEniCS/dolfinx/blob/main/.clang-format