Vaishnav Tadiparthi

Would this be a MWE? ```julia using CUDA, Flux, Zygote CUDA.allowscalar(false) dim = 2; nn = 10; chain = Chain(Dense(dim, nn, tanh), Dense(nn, nn, tanh), Dense(nn, 1)); θ, re =...

Thanks for the quick response! So I am able to register the functions inside `f(x)` that are not symbolically representable, but those inner functions require derivatives too. Assume that in...

Okay, a portion of my `f(x)` looks like this: ```julia f([xV, xα]) = xV + 0.0031009487651181677normV(xV)*cos(xα) ``` normV(xV) has been `@register`-ed. Simplifying my problem for the sake of explanation, the...

So AD isn't called by default when parsing through `@register`-ed function expressions which require derivatives? The AD derivatives of such functions have to be manually created and registered for MTK...

Please let me know if this is not a NeuralPDE issue but a MTK or a Symbolics issue.

Check [here](https://gist.github.com/vaishnavtv/3d6ba4748c4f9ac389240d8167b0fcb2). I've attached both `Project.toml` and `Manifest.toml`.

When I run `rm MechanismGeometries` and `rm MeshCat` to simply install MeshCatMechanisms, I'm getting this error: ERROR: Unsatisfiable requirements detected for package MeshCat [283c5d60]: MeshCat [283c5d60] log: ├─possible versions are:...

Um, I was messing around with the packages, and somehow I've been able to install the package. I don't know what exactly changed, but I suppose it's working for now....

So, as I understand, the problem seems to disappear if I add the package from my project folder, but persists if I add it directly from my REPL. Does that...

No, I'm sorry that wasn't clear. When I create a virtual environment using `Pkg.activate(@__DIR__)`, I'm able to add MeshCatMechanisms, but if I do this in the global(?) JuliaPro environment, I...