Variational preparation of finite-temperature states on a quantum computer

TL;DR

This paper introduces a variational method to prepare finite-temperature states on quantum computers, demonstrated experimentally with high fidelity, enabling more accurate quantum simulations of thermal systems.

Contribution

It presents a novel hybrid quantum-classical variational approach to generate thermofield double states without prior parameter optimization, validated on a real quantum processor.

Findings

Achieved state fidelities from 99% to 75% across temperature range

Experimental results agree with numerical simulations considering noise

Method enables efficient thermal state preparation on near-term quantum devices

Abstract

The preparation of thermal equilibrium states is important for the simulation of condensed-matter and cosmology systems using a quantum computer. We present a method to prepare such mixed states with unitary operators, and demonstrate this technique experimentally using a gate-based quantum processor. Our method targets the generation of thermofield double states using a hybrid quantum-classical variational approach motivated by quantum-approximate optimization algorithms, without prior calculation of optimal variational parameters by numerical simulation. The fidelity of generated states to the thermal-equilibrium state smoothly varies from 99 to 75% between infinite and near-zero simulated temperature, in quantitative agreement with numerical simulations of the noisy quantum processor with error parameters drawn from experiment.