Tensor networks are never defined

[gdalle]

Although I am familiar with the literature on graphical models, I didn't know what a tensor network was, or why the contraction order is important. I had to look up TensorOperations.jl, which in turn redirected me to this paper before I could understand this crucial aspect.

Both in the docs and in the JOSS paper, there needs to be an introduction to tensor networks and why they're harder to handle than it appears

https://github.com/openjournals/joss-reviews/issues/5700

[mroavi]

Absolutely, glad you pointed that out. We've added a whole new section in the documentation to introduce tensor networks and explain their connection with probabilistic graphical models. You can check it out here: https://tensorbfs.github.io/TensorInference.jl/dev/tensornetwork/. As for the JOSS paper, we've reached our word limit. Do you have any advice on how to include this new information without exceeding (more) the word count?

[gdalle]

Cool addition! I'll give some thought to what we can trim from the paper

[mroavi]

The paper now includes a short introduction to tensors, contractions, and tensor networks, and it explains why the contraction order is important.

[GiggleLiu]

NOTE: The above link to tensor network definition has been updated. https://tensorbfs.github.io/TensorInference.jl/dev/tensor-networks/

[gdalle]

great!