The Localized Active Space Method with Unitary Selective Coupled Cluster

TL;DR

This paper presents LAS-USCCSD, a hybrid quantum-classical method that reduces quantum circuit complexity by selectively identifying important parameters, enabling more practical multireference calculations on near-term quantum computers.

Contribution

LAS-USCCSD introduces a parameter selection strategy that significantly decreases circuit depth compared to LAS-UCCSD, improving feasibility for near-term quantum applications.

Findings

LAS-USCCSD reduces parameters and circuit depth by at least tenfold.

Benchmark results show accurate energies for small molecules and magnetic properties.

Method enhances practicality of multireference quantum algorithms on near-term devices.

Abstract

We introduce a hybrid quantum-classical algorithm, the localized active space unitary selective coupled cluster singles and doubles (LAS-USCCSD) method. Derived from the localized active space unitary coupled cluster (LAS-UCCSD) method, LAS-USCCSD first performs a classical LASSCF calculation, then selectively identifies the most important parameters (cluster amplitudes used to build the multireference UCC ansatz) for restoring inter-fragment interaction energy using this reduced set of parameters with the variational quantum eigensolver method. We benchmark LAS-USCCSD against LAS-UCCSD by calculating the total energies of $(\mathrm{H}_2)_2$, $(\mathrm{H}_2)_4$ and \textit{trans}-butadiene, and the magnetic coupling constant for a bimetallic compound [Cr$_2$(OH)$_3$(NH$_3$)$_6$]$^{3+}$. For these systems, we find that LAS-USCCSD reduces the number of required parameters and thus the circuit depth by at least one order of magnitude, an aspect which is important for the practical implementation of multireference hybrid quantum-classical algorithms like LAS-UCCSD on near-term quantum computers.