# A simple method for seniority-zero quantum state preparation

## TL;DR

This paper introduces a shallow-circuit method for preparing high-fidelity singlet states using a simplified pCCD approach, facilitating quantum phase estimation of strongly correlated electronic systems.

## Contribution

It proposes a novel, efficient state preparation technique by substituting leading oo-pCCD amplitudes into UpCCD, enabling practical quantum simulations of complex molecules.

## Key findings

- High-fidelity singlet states achieved with shallow circuits
- Effective for models of multiple-bond dissociation and Hubbard models
- Simplifies initial state preparation for quantum algorithms

## Abstract

Quantum Phase Estimation (QPE), the quantum algorithm for estimating eigenvalues of a given Hermitian matrix and preparing its eigenvectors, is considered the most promising approach to finding the ground states and their energies of electronic systems using a quantum computer. It requires, however, to be warm--started from an initial state with sufficiently high overlap with the ground state. For strongly-correlated states, where QPE is expected to have advantage over classical methods, preparation of such initial states requires deep quantum circuits and/or expensive hybrid quantum-classical optimization. It is well-known that orbital-optimized paired Coupled Cluster Doubles (oo-pCCD) method can describe the static correlation features of many strongly correlated singlet states. We show that pCCD and its unitary counterpart, UpCCD, become equivalent in the limit of small amplitudes or if the number of large amplitudes is below 5. We demonstrate that substituting leading oo-pCCD amplitudes into the UpCCD Ansatz allows to prepare high-fidelity singlet states for models of multiple-bond dissociation in ethene, ethyne and dinitrogen, as well as for 1D Hubbard models at half-filling, with very shallow circuits. We envisage our method to be of general use for approximate preparation of singlet states for Quantum Phase Estimation and related algorithms.

---
Source: https://tomesphere.com/paper/2508.21679