[ITensors] [BUG] Regression introduced to ITensors during refactor of NDTensors away from kwargs...

[brian-dellabetta]

Description of bug

A recent change to NDTensors introduced a regression into ITensors' replacebond! function for mps, see here for diff and responsible call to factorize just underneath. Previously, users could set min_blockdim, use_absolute_cutoff, and use_relative_cutoff in the keyword args passed to replacebond!, and kwargs... would be passed correctly to factorize. This refactor away from kwargs is 100% welcomed as it makes the API more clear and readable, but it seems to have missed those fields, and possibly others or in other functions?

It also appears that factorize currently doesn't pass svd-related fields correctly into factorize_svd (see here, the svd fields don't seemed to be used anywhere in fact.

Happy to create a PR to add those fields if you agree, just let me know.

Minimal code demonstrating the bug or unexpected behavior

This succeeds in ITensors v0.3.48 & NDTensors v0.2.15
But fails in ITensors v0.3.52 & NDTensors v0.2.23:

using ITensors
mps=randomMPS(Float64, [Index(2) for _ in 1:5])
replacebond!(mps, 2, mps[3]; which_decomp="svd", use_relative_cutoff=true, min_blockdim=1)

Error in v0.3.52:

ERROR: MethodError: no method matching replacebond!(::MPS, ::Int64, ::ITensor; which_decomp::String, use_relative_cutoff::Bool, min_blockdim::Int64)

Closest candidates are:
  replacebond!(::MPS, ::Int64, ::ITensor; normalize, swapsites, ortho, which_decomp, mindim, maxdim, cutoff, eigen_perturbation, svd_alg) got unsupported keyword arguments "use_relative_cutoff", "min_blockdim"

Version information

• Output from versioninfo():

julia> versioninfo()
Julia Version 1.10.0
Commit 3120989f39b (2023-12-25 18:01 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: macOS (x86_64-apple-darwin22.4.0)
  CPU: 12 × Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, skylake)
  Threads: 1 on 12 virtual cores

[mtfishman]

Thanks for the report, I'm not surprised some keyword arguments were lost in that PR since it was a pretty big undertaking. A PR would be welcome!

[brian-dellabetta]

#1300 submitted to resolve this issue, please lmk what you think!