Improved Lanczos Algorithm using Matrix Product States (2504.21786v1)

Abstract: We improve the Lanczos algorithm using the matrix product state representation proposed in Phys. Rev. B 85, 205119 (2012). As an alternative to the density matrix renormalization group (DMRG), the Lanczos algorithm avoids local minima and can directly find multiple low-lying eigenstates. However, its performance and accuracy are affected by the truncation required to maintain the efficiency of the tensor network representation. In this work, we enhance its convergence by restarting with multiple states. We benchmark our method on one-dimensional instances of the Fermi-Hubbard model with 8 sites and the Heisenberg model with 16 sites in an external field, using numerical experiments targeting the first five lowest eigenstates. Across these tests, our approach obtains accuracy improvements of three to seven orders of magnitude. Finally, we extend the Heisenberg model simulation to a lattice with 30 sites to highlight its scalability.