Error estimates for extrapolations with matrix-product states

TL;DR

This paper introduces a new, computationally efficient error measure for matrix-product states that enables accurate extrapolation of ground state properties without the high cost of traditional methods.

Contribution

A novel error measure based on an approximation of the full variance that does not require two-site DMRG, facilitating efficient and reliable extrapolation in MPS calculations.

Findings

The new error measure correlates well with traditional measures.

Extrapolation using the new error measure matches the accuracy of more costly methods.

The approach is demonstrated on multiple quantum many-body models.

Abstract

We introduce a new error measure for matrix-product states without requiring the relatively costly two-site density matrix renormalization group (2DMRG). This error measure is based on an approximation of the full variance $\langle \psi | ( \hat H - E )^2 |\psi \rangle$. When applied to a series of matrix-product states at different bond dimensions obtained from a single-site density matrix renormalization group (1DMRG) calculation, it allows for the extrapolation of observables towards the zero-error case representing the exact ground state of the system. The calculation of the error measure is split into a sequential part of cost equivalent to two calculations of $\langle \psi | \hat H | \psi \rangle$ and a trivially parallelized part scaling like a single operator application in 2DMRG. The reliability of the new error measure is demonstrated at four examples: the $L=30, S=\frac{1}{2}$ Heisenberg chain, the $L=50$ Hubbard chain, an electronic model with long-range Coulomb-like interactions and the Hubbard model on a cylinder of size $10 \times 4$. Extrapolation in the new error measure is shown to be on-par with extrapolation in the 2DMRG truncation error or the full variance $\langle \psi | ( \hat H - E )^2 |\psi \rangle$ at a fraction of the computational effort.