Quantum resource reduction for quantum-centric supercomputing via correlated mean-field downfolding framework

TL;DR

OBDF-SQD is a hybrid quantum-classical method that reduces quantum resource requirements by combining classical downfolding with quantum sampling, improving accuracy over traditional active-space methods.

Contribution

This work introduces OBDF-SQD, a novel hybrid approach that integrates one-body downfolding with quantum diagonalization to enhance quantum-centric supercomputing efficiency.

Findings

OBDF-SQD outperforms CAS-SQD in benchmark tests.

The method retains the same operator structure, avoiding extra quantum circuit complexity.

It is extendable to periodic solids within quantum embedding frameworks.

Abstract

We present OBDF-SQD, a hybrid quantum-classical method that combines one-body downfolding~(OBDF) based on one-body M\o{}ller--Plesset second-order perturbation theory (OBMP2) with sample-based quantum diagonalization~(SQD) for use in quantum-centric supercomputing~(QCS). In this approach, OBMP2 is executed classically to fold dynamical correlation from external orbitals into a renormalized one-body operator, yielding an effective active-space Hamiltonian that retains the same operator structure as the bare Hamiltonian and therefore requires no additional quantum circuit resources. SQD is then applied to this effective Hamiltonian, where, in this work, the quantum sampling is performed via the Qiskit Aer simulator rather than actual quantum hardware. We benchmark OBDF-SQD on dissociation curves of \ce{H6} chain, ring, and lattice systems and the \ce{N2} molecule in the cc-pVDZ basis, comparing against standard methods and active-space SQD (CAS-SQD). We observed that OBDF-SQD consistently improves upon CAS-SQD with the same active space. The simplicity of the one-body downfolding correction also makes the approach straightforwardly extensible to periodic solids within existing quantum embedding frameworks