Matt Fishman

Yes, I just meant that it would be breaking from the current behavior, so we should decide on it pre-0.1. Agreed that we can add features like `permute` for `Diag`...

Are we going to fix the scaling of `correlation_matrix(::MPO, ...)` in this PR or leave that for a future PR?

Also on a related note, it would be nice to have a function that acts like `removeqns` but within `NDTensors`, perhaps called `removeblocks`. This would do the conversions: - `BlockSparse...

The reason for this is because essentially this is implemented using a staircase circuit structure, which causes the asymmetry in the bond dimension. You should be able to use the...

Great, looks like you found a simple fix using the current code in #870.

Sounds good. However with the multi-operator notation: ```julia expect(psi,"Sz",j,"Sx",j,"Sz",k) ``` my first reaction was that it looks like it would be calculating a correlation function ``. I would suggest something...

Yeah, in that case it seems like it could be a minimal amount of type information to reconstruct the original object. In general it isn't necessary to store `Index{Int}` since...

Thanks, this seems reasonable. Two comments: 1. Should we define a special compact printing for when the Indices are in a collection, like is being proposed for MPS printing in...

One issue with the plan above is that the way to define the compact version of showing an object is by overloading `show(::IO, ::Indices)` which is what `@show` uses, so...

Maybe the `compact` attribute of `IO` could be used for this? See: https://discourse.julialang.org/t/show-and-showcompact-on-custom-types/8493 https://github.com/JuliaLang/julia/blob/9028130434c5fb55e00c65e73aaf13cf0afde380/base/complex.jl#L181-L196