Calculating the ground state energy of benzene under spatial deformations with noisy quantum computing

TL;DR

This study evaluates the feasibility of using noisy quantum computers with variational algorithms to calculate the ground state energy of benzene under deformation, highlighting the strengths and limitations of different ansatze.

Contribution

It compares hardware efficient and qUCC ansatze for quantum chemistry on near-term devices, analyzing noise effects and optimization challenges in practical quantum computations.

Findings

Hardware efficient ansatz can outperform mean-field methods under extreme deformations.

Results depend heavily on initial parameters and optimization, especially in noisy conditions.

qUCC ansatz is more sensitive to noise and less effective with current hardware limitations.

Abstract

In this manuscript, we calculate the ground state energy of benzene under spatial deformations by using the variational quantum eigensolver (VQE). The primary goal of the study is estimating the feasibility of using quantum computing ansatze on near-term devices for solving problems with large number of orbitals in regions where classical methods are known to fail. Furthermore, by combining our advanced simulation platform with real quantum computers, we provided an analysis of how the noise, inherent to quantum computers, affects the results. The centers of our study are the hardware efficient and quantum unitary coupled cluster ansatze (qUCC). First, we find that the hardware efficient ansatz has the potential to outperform mean-field methods for extreme deformations of benzene. However, key problems remain at equilibrium, preventing real chemical application. Moreover, the hardware efficient ansatz yields results that strongly depend on the initial guess of parameters - both in the noisy and noiseless cases - and optimization issues have a higher impact on their convergence than noise. This is confirmed by comparison with real quantum computing experiments. On the other hand, the qUCC ansatz alternative exhibits deeper circuits. Therefore, noise effects increase and are so extreme that the method never outperform mean-field theories. Our dual simulator/8-16 qubits QPU computations of qUCC appears to be a lot more sensitive to hardware noise than shot noise, which give further indications about where the noise-reduction efforts should be directed towards. Finally, the study shows that qUCC method better captures the physics of the system as the qUCC method can be utilized together with the Huckel approximation. We discussed how going beyond this approximation sharply increases the optimization complexity of such a difficult problem.