maartenvd

This has been a real pain in the past, so at some point I decided to switch over to the slower-but-stable method in mpskit (at some point we should probably...

I hoped gesdd simply failed to converge, and that the data would be salvageable. If it isn't, then the overhead of a copy combined with the slow garbage collector might...

so plan of action - we first change tensorkitmanifolds to allow passing in the svd algorithm, and then change mpskit to pick the stable one? Op do 25 jan 2024...

To tag on to this, should we change the approximate call signature to not take in tuples? I'm not sure why I started doing that, but at the moment it...

While it's possible that lazysum gives the incorrect gradient, I would've expected gradientgrassmann to then completely deadlock, as optimkit's linesearch is quite picky about the correctness of the gradient. LazySum...

> @maartenvd , if I am correct, the motivation for the current rather complicated implementation is that, if you just use the standard gradient of the left canonical form, and...

I think I started with virtualspace referring to the right virtualspace. However in hindsight, it is more natural to make this refer to the left virtualspace instead. If you write...

for C we have CL and CR, so that AL(n) - CR(n) = CL(n) - AR(n). They're just shifted copies of eachother, but maybe we should pick a convention and...

optimkit needs to be given a gradient and a descent direction (in gradient descent you would make the descent direction be - gradient). A descent direction *has* to satsify that...

It may be worth mentioning that the paper you linked is not directly related to the excitation ansatz as it is typically used. The paper deals with superpositions of operators...