Understanding finite size effects in quasi-long-range orders for exactly solvable chain models

TL;DR

This paper examines how finite size effects influence the numerical analysis of quasi-long-range order in exactly solvable chain models, highlighting boundary condition impacts and finite size scaling limitations.

Contribution

It provides a detailed comparison of finite-size effects under different boundary conditions in exactly solvable models, revealing systematic biases and convergence behaviors.

Findings

Finite-size scaling can partially remove numerical artifacts.

Boundary conditions significantly affect correlation function exponents.

Convergence to the infinite limit occurs at chain lengths > 2000.

Abstract

In this paper, we investigate how much of the numerical artefacts introduced by finite system size and choice of boundary conditions can be removed by finite size scaling, for strongly-correlated systems with quasi-long-range order. Starting from the exact ground-state wave functions of hardcore bosons and spinless fermions with infinite nearest-neighbor repulsion on finite periodic chains and finite open chains, we compute the two-point, density-density, and pair-pair correlation functions, and fit these to various asymptotic power laws. Comparing the finite-periodic-chain and finite-openchain correlations with their infinite-chain counterparts, we find reasonable agreement among them for the power-law amplitudes and exponents, but poor agreement for the phase shifts. More importantly, for chain lengths on the order of 100, we find our finite-open-chain calculation overestimates some infinite-chain exponents (as did a recent density-matrix renormalization-group (DMRG) calculation on finite smooth chains), whereas our finite-periodic-chain calculation underestimates these exponents. We attribute this systematic difference to the different choice of boundary conditions. Eventually, both finite-chain exponents approach the infinite-chain limit: by a chain length of 1000 for periodic chains, and > 2000 for open chains. There is, howwever, a misleading apparent finite size scaling convergence at shorter chain lengths, for both our finite-chain exponents, as well as the finite-smooth-chain exponents. Implications of this observation are discussed.