Matt Fishman

Definitely makes sense to avoid setting indices/changing tags when it isn't necessary, so your proposal makes sense to me to do anyway. Maybe we have to think about the copying...

That sounds reasonable, @b-kloss does that make sense to you? A slightly simpler way to write it would be: ```julia function orderings(::TDVPOrder{4,direction}) where {direction} return repeat(orderings(TDVPOrder{2,direction}(), 3) end ```

You can make a fork and then a PR, no need for access rights.

That's correct, it just uses a dense diagonalization right now. I would say that my goal for `dmrg_x` was to present the simplest demonstration of a standard method, and not...

@emstoudenmire https://github.com/mtfishman/ITensorNetworks.jl/pull/43 is now merged so this can be moved to that repository.

@emstoudenmire seems like people are itching to use this, but that it became outdated since it is relying on an internal ITensors.jl function (`ITensors.directsum_itensors`) that was removed in https://github.com/ITensor/ITensors.jl/pull/1185. Should...

Something I see that is missing is expanding by a sum of Hamiltonians, I guess that would be easy enough to add on later.

@emstoudenmire is there a way to organize the `extend` method such that it first creates an MPS `psi_extend` that stores the expansion on each site, and then uses `+(psi, psi_extend;...

@emstoudenmire I've done a pretty extensive rewrite, though it is mostly aesthetic changes and the basic algorithmic structure remains the same. I've also proposed a new interface, let me know...

@emstoudenmire in the latest I changed `expand_basis(...)` to `expand(...)`. I think it is more "punchy", hopefully it is not too terse but I think it should be clear what it...