[Yue-Zhengyuan]

This PR extends add_physical_charge to work with MPOTerm and TensorProductTerm of a LocalOperator introduced in #425.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
src/operators/localoperator.jl	84.76% <ø> (ø)
src/operators/localoperator/mpo.jl	100.00% <100.00%> (ø)
src/operators/localoperator/tensorproduct.jl	95.65% <100.00%> (-4.35%)	⬇️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[leburgel]

This seems great, I'll have a closer look later today. I did have one question right away.

The MPOTerm _fuse_charge implementation now has a very big overlap with what MPSKit.add_physical_charge does within MPSKit itself. Do you think there's any opportunity to reuse shared functionality from there? We don't have to forcibly fit this into the MPSKit implementation if it doesn't make sense, but since they really do the same thing it might make sense to share core routines.

This can also be done afterwards with some changes on both sides if it's not natural with the current state.

[Yue-Zhengyuan]

To reuse MPSKit, the result looks like this:

function _fuse_mpo_charge(tensor::AbstractTensorMap, charge::Sector)
    nout, nin = numout(tensor), numin(tensor)
    # MPSKit expects both virtual legs, including trivial legs at MPO endpoints.
    nout == 1 && (tensor = insertleftunit(tensor, 1))
    nin == 1 && (tensor = insertrightunit(tensor, numind(tensor)))
    # MPSKit dualizes its charge argument; LocalOperator has already done so.
    tensor = MPSKit.add_physical_charge(tensor, dual(charge))
    nin == 1 && (tensor = removeunit(tensor, 4))
    nout == 1 && (tensor = removeunit(tensor, 1))
    return tensor
end

The extra handling does not cost much, but this does remind me if we want to support PBC-MPOTerms.