Entanglement entropy of any subsystem

Hello All,

I have a question regarding calculating the entanglement entropy (EE) of a subregion. Let’s say I want to calculate EE of subsytstem A for case A and B shown below.


For case A, this calculation of EE is carried out by using ‘orthogonalize’ and ‘svd’. The example is shown here: (MPS and MPO Examples · ITensors.jl)
In case B where subsystem A is relocated to the middle of the chain. My question is: how do I compute the EE in this case ? Is ‘svd’ still viable in this case ? I hope someone can enlight me in some way. Thank you !

Hi, you can use for instance the approach posted by the OP here - Is this a correct way to calculate one-site and two-site entanglement spectrum and entropy?. Then from the eigenvals. of the reduced density matrix you can easily get the entanglement entropy and other such quantities.

This works very well as long as the subsystem isn’t very large (a few sites or so, and also depends on the number of indices involved).