Efficient approximation for global functions of matrix product operators
This work provides a more efficient approach for computing spectral functions in tensor network methods, benefiting researchers studying quantum many-body systems.
The paper presents an efficient method to approximate operator functions of Hermitian matrix product operators, enabling computation of spectral quantities like von Neumann entropy and trace norm. The method shows improved performance for thermal properties of Hamiltonians, demonstrated on Lipkin-Meshkov-Glick and Ising models.
Building on a previously introduced block Lanczos method, we demonstrate how to approximate any operator function of the form Trf (A) when the argument A is given as a Hermitian matrix product operator. This gives access to quantities that, depending on the full spectrum, are difficult to access for standard tensor network techniques, such as the von Neumann entropy and the trace norm of an MPO. We present a modified, more efficient strategy for computing thermal properties of short- or long-range Hamiltonians, and illustrate the performance of the method with numerical results for the thermal equilibrium states of the Lipkin-Meshkov-Glick and Ising Hamiltonians.