Jutho

Hi @amilsted. My apologies for the late response; I was on holidays. This is probably a good suggestion. I initially developed the CUDA wrappers in a separate package (https://github.com/Jutho/CuTensorOperations.jl ),...

I am certainly open to suggestions, but how would you like this to work or look like, even for the case of a tensor without symmetries (i.e. an object of...

Maybe the construction with the embedding is actually the easiest way also to provide an actual implementation. I often find the most difficult question to be the name and interface...

What `data1` and `data2` do you have in mind? I.e. could you just post a runnable example, using some random generated data, so that I can test what is going...

I can reproduce your timings and observe that something is going wrong with threading. When printing out the `threadid()` when actually doing the work, it shows `1` is most cases,...

Although, digging a bit deeper, I don't think that's really the issue. Giving the large number of allocations, there seems to be something going on with type inference. Note that...

I guess this has to do with some compiler heuristics on the complexity of tuples and nested parametric types.

There is a constant `const MINTHREADLENGTH = 1

I have not been following very actively myself all the recent developments. I think you want to check out things like LoopVectorization.jl , which will also do other optimizations

The tutorial can probably use some work. This notation means, build a vector space which carries a representation of ℤ₂ where irrep 0 (the trivial irrep) appears 3 times, and...