Evaluating spectral functions with matrix product states
Evaluating spectral functions with matrix product states
March 9, 2022 (Wed.) at 1:30PM (ET)
March 9, 2022 (Wed.) at 1:30PM (ET)
University of Gent, Belgium
University of Gent, Belgium
The spectral function is one of the most important quantities that link theory and experiment in the field of strongly-correlated quantum matter, but its numerical evaluation for a given microscopic model is a notoriously hard problem. For one-dimensional systems, the formalism of matrix product states (MPS) has been instrumental for obtaining accurate spectral functions. These MPS techniques can be used in two dimensions by placing the system on a cylinder, but this requires huge computational resources for getting accurate results on relatively narrow cylinders. In this talk, we show how the efficiency of MPS methods can be improved by evaluating spectral functions directly in momentum space. We show that we can simulate the time evolution after applying a momentum operator to the ground state of an infinite (quasi) 1-D system, and that the entanglement growth is considerably smaller than after applying a real-space operator. This allows us to simulate the time evolution to much longer times with the same computational cost, which leads to much more precise spectral functions after transforming to frequency space. We show applications of these state-of-the-art MPS methods for spectral functions of one- and two-dimensional models with fractionalized excitations.