TDVP for density matrix and local observable expectation value

DMRG and Numerical Methods
1

I am new to itensor and would like to use it for my problem but got stuck in how to write an efficient code. I want to time evolve a mixed state using a Hamiltonian and then measure some local observables. I have two questions:

  1. Is there a way to apply TDVP for density matrix? Are there some examples?

  2. For local observable O expectation value, I was using inner(rho, O) where O is constructed as a MPO. It seemed to me this is not efficient, since O only acts on one or two sites while the MPO is constructed on all sites. Is there a more efficient way of doing this?

2

As for as I know, ITensorMPS.jl does not support mixed state calculations.

I recommend using TensorMixedStates.jl. This package can handle time evolution based on Schrodinger and Lindblad equations, and is able to calculate expectations for density matrix efficiently.

You can check this: https://arxiv.org/pdf/2505.11377

Best

1 Like
3

Thanks a lot! This is very helpful.