Tensor decompositions in Julia

[00krishna]

I was wondering if there is a package for tensor decomposition was in Julia–hopefully the package is differentiable as well for ML. I am talking about CP-decompositions and Tucker decompositions to start. I actually would like to try some tensor regression as well. I was thinking of something akin to the Tensorly project in python if that helps.

I found a few posts on this topic, but they are pretty old.

One user used the TensorDecompositions.jl package, but that has not been updated in 6 years.

Low rank tensor computation

I want to generate a rank-1 tensor with Julia for some tensor decomposition. As for SVD and linear combination of singular values and orthogonal basis vectors, one can decompose a matrix by a sum like the sum of singular values * u * v’. I can generate a 10x20x30 matrix like this. (the original code is inspired by GitHub - yunjhongwu/TensorDecompositions.jl: A Julia implementation of tensor decomposition algorithms) u = randn(10); v = randn(20); w = randn(30) # Generate a noisy rank-1 tensor…

Tullio.jl seems good, but then it seems I would need to code the tensor decomposition myself. That is fine, but since I am new to tensor decompositions it would be nice to have a reference implementation to compare against.

Finally, the TensorOperations.jl package indicates that it is for tensor contractions-similar to Tullio, however I did not see any reference to tensor decomposition, etc.