Avoid in-place operations in adjoint and conj functions

[jofrevalles]

Summary

This PR modifies the adjoint and conj functions for AbstractQuantum and AbstractTensorNetwork types to prevent errors when tensors contain immutable arrays. Instead of performing in-place operations, the functions now return a modified copy of the input.

Some tensors within a TensorNetwork may have immutable arrays in their parent structures. Performing in-place operations like conj! or adjoint! on these tensors can lead to errors because immutable arrays do not support the setindex! operation required for in-place modification.

Example

julia> using Tenet

julia> using YaoBlocks

julia> c = YaoBlocks.chain(2)

julia> c = YaoBlocks.chain(2, c, put(1 => X))

julia> c = YaoBlocks.chain(2, c, put(2 => X))

julia> c = YaoBlocks.chain(2, c, put(1 => X))

julia> allzeros = Quantum(Product(fill([1, 0], 2)))
Quantum (inputs=0, outputs=2)

julia> state_qtn = merge(allzeros, Quantum(c))
Quantum (inputs=0, outputs=2)

Before this PR:

julia> state_qtn'
ERROR: BoundsError: attempt to access 2×2 LuxurySparse.SDPermMatrix{ComplexF64, Int64, Vector{ComplexF64}, Vector{Int64}} at index [1, 1]
Stacktrace: ...

Now:

julia> state_qtn'
Quantum (inputs=2, outputs=0)

[mofeing]

Are there already tests for adjoint and conj? I guess that yes but not sure right now.

[jofrevalles]

Are there already tests for adjoint and conj? I guess that yes but not sure right now.

I am writting these right now.

[jofrevalles]

By the way, these errors appear now due to the issue reported in #225, since when we do conj now we call replace! inside, and that errors when the arrays of the tensors are Real, since then conj(tensor)=tensor.

See:

julia> mps = rand(Chain, Open, State, n=6, χ=10)
MPS (inputs=0, outputs=6)

julia> mps'
ERROR: Index A must be open
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] addsite!(tn::Quantum, site::Site{1}, index::Symbol)
   @ Tenet ~/git/Tenet.jl/src/Quantum.jl:279
 [3] adjoint(tn::Quantum)
   @ Main ./REPL[57]:8
 [4] adjoint(chain::Chain)
   @ Tenet ~/git/Tenet.jl/src/Ansatz/Chain.jl:209
 [5] top-level scope
   @ REPL[74]:1

julia> mps = rand(Chain, Open, State, n=6, χ=10, eltype=ComplexF64)
MPS (inputs=0, outputs=6)

julia> mps'
MPS (inputs=6, outputs=0)

[jofrevalles]

When the PR #227 is merged, I will cherry-pick the fixing commits to ensure everything works well.

[mofeing]

When the PR #227 is merged, I will cherry-pick the fixing commits to ensure everything works well.

You don't need to cherry-pick if the PR is merged??

[jofrevalles]

When the PR #227 is merged, I will cherry-pick the fixing commits to ensure everything works well.

You don't need to cherry-pick if the PR is merged??

Just to run the tests before merge.

[mofeing]

Ah, you don't need to cherry-pick but to rebase on top of master.

Actually, cherry-picking in that case can lead to merge conflicts.

[mofeing]

@jofrevalles tests should be solved by just rebasing

git checkout fix/inplace-conj
git rebase master
git push --force-with-lease