maartenvd

I feel rather silly, as the second example is actually also an error. ``` temp = call_matlab(1,"feval",[args[1],temp]) ``` call_matlab will return a vector, which gets put into temp. temp then...

Ok so the problem is that my R is rank-defficient, and the inverse (which is done using trtrs) fails. I made sure my R's are now full rank and everything...

>LinearAlgebra backward functions have the problem of exploding gradients. When do they have this problem? I know svd has it, but I assumed that the backward differentiation of qr was...

There is no "real" support for linbladians, in the sense that all implemented operators (mpo/hamiltonian) act on MPS tensors with one physical leg. You can either fuse the two physical...

I haven't investigated it further, but I see a significant (10x) performance (1ms to 8ms) decrease when switching from v0.12.7 to this branch. I also expected having to do even...

I completely forgot that the underlying data field also changed! The speedup would come from the re-ordered fusiontrees, but I'll need to change the code a little bit

I was testing an old version of jh/tensorstructure (with some extra local changes). I will reimplement on the latest master today, it really should be strictly faster (less matmuls and...

yep - will also check if the blockstructure caching gives a sufficient enough speedup to do away with some of the uglier things in mpskitexperimental/qchem, but that will take more...

My updated submult is even more of a mess but - it now indeed only does 27 mul! calls, instead of 281 - it is only a little bit slower...

Also, if recurse!(::Nothing) was defined, couldn't the dynamo example be even prettier: ``` @dynamo function quicktest(a...) ir = IR(a...) recurse!(ir) return ir end ```