Lukas Devos
Lukas Devos
Almost working version of the eigsolve rrules. Seems to be failing only for real matrices, where comparison with rrule for eigen does not give equal results. Specifically, it seems that...
rewrite algorithms in terms of VectorInterface methods. * Some additional cleanup can be done in terms of scalartype * algorithms can be written in terms of bangbang-methods, possibly allowing tuples...
Sometimes, it might be useful to have the option to start with multiple initial vectors. In particular, when the first 2 dominant eigenvectors are known, but one would like to...
Uses `SoftGlobalScope` to mimic IJulia behaviour for notebooks. This eliminates the need of `global` in loops at top level. Optionally, this could be enabled for .jmd files as well, allthough...
Noted by @leburgel: When contracting a network that results in a scalar, currently `ncon` does not insert a final call to `tensorscalar`. As such the following (unintuitive) behaviour is happening:...
Adds a non-unicode alternative to `\otimes` for sectors: `fusionproduct`. Fixes #108 .
Major overhaul of too many things, will split into parts but want to run tests first
Having a look at the output of the tests, it seems like there is some performance issue going on with the combination of `LazySum` and `GradientGrassmann`. My best guess is...
When starting with an InfiniteMPS from tensors that are not full rank, the gauging algorithm does not necessarily lead to square `CR` tensors. If this happens at the edges of...