Perturbative Variational Quantum Eigensolver via Reduced Density Matrices

Huili Zhang; Jie Liu; Jinlong Yang; Xiaoxia Cai; Xiongzhi Zeng; Yibin Guo; Yuhan Zheng; Zhenyu Li

arxiv: 2504.02340 · v2 · submitted 2025-04-03 · 🪐 quant-ph

Perturbative Variational Quantum Eigensolver via Reduced Density Matrices

Yuhan Zheng , Yibin Guo , Huili Zhang , Jie Liu , Xiongzhi Zeng , Xiaoxia Cai , Zhenyu Li , Jinlong Yang This is my paper

Pith reviewed 2026-05-22 21:27 UTC · model grok-4.3

classification 🪐 quant-ph

keywords variational quantum eigensolverperturbation theoryreduced density matricesquantum chemistryactive spacedynamic correlationpotential energy surfaceNISQ

0 comments

The pith

Augmenting variational quantum eigensolver with perturbative correction from reduced density matrices captures dynamic correlation beyond the active space without added qubits or circuit depth.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a hybrid method that solves the effective Hamiltonian for a compact active space using the variational quantum eigensolver and then computes a perturbative energy correction from the reduced density matrices of that solution. This correction accounts for dynamic electron correlation outside the active space while leaving the quantum circuit unchanged. A sympathetic reader would care because current quantum devices have too few high-fidelity qubits and short coherence times to simulate full molecular electron correlation directly. Numerical tests on HF and N2 show systematic gains over plain active-space VQE, and an experiment on F2 reaches a mean absolute error of 1.2 millihartree along the potential energy surface when combined with error mitigation.

Core claim

The effective Hamiltonian in the active space is solved by VQE, and the perturbative energy correction is computed from reduced density matrices, thereby avoiding any increase in circuit depth or qubit overhead. Numerical simulations on HF and N2 demonstrate systematic improvements over standard VQE within compact active spaces. An experimental realization for F2 on a superconducting quantum processor, in conjunction with robust error mitigation strategies, achieves a mean absolute error of 1.2 millihartree along the potential energy surface.

What carries the argument

The perturbative correction to the active-space energy, computed from the reduced density matrices obtained from the VQE solution.

Load-bearing premise

The perturbative energy correction computed from the reduced density matrices of the active-space VQE solution accurately captures the dynamic electron correlation effects outside the chosen compact active space for the tested molecules.

What would settle it

Full configuration interaction calculations for HF and N2 showing that the added perturbative corrections do not reduce the energy error relative to active-space VQE results would falsify the improvement claim.

read the original abstract

Current noisy intermediate-scale quantum (NISQ) devices remain limited in their ability to perform accurate quantum chemistry simulations due to restricted numbers of high-fidelity qubits and short coherence times. To overcome these challenges, we introduce the perturbative variational quantum eigensolver (VQE-PT), a hybrid quantum-classical algorithm that augments VQE with perturbation theory to account for electron correlation effects beyond a compact active space. Within this framework, the effective Hamiltonian in the active space is solved by VQE, and the perturbative energy correction is computed from reduced density matrices, thereby avoiding any increase in circuit depth or qubit overhead. We benchmark the proposed algorithm through numerical simulations on HF and N$_2$, demonstrating systematic improvements over standard VQE within compact active spaces. Furthermore, we perform an experimental realization on the Quafu superconducting quantum processor for $\rm F_2$, where, in conjunction with robust error mitigation strategies, the method achieves high accuracy (a mean absolute error of 1.2 millihartree) along the potential energy surface. These results demonstrate VQE-PT as a practical and resource-efficient pathway for incorporating dynamic correlation in quantum chemistry simulations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

VQE-PT computes a perturbative correction from active-space VQE RDMs to add dynamic correlation without extra qubits or depth, with simulations on HF/N2 and a Quafu F2 experiment claiming 1.2 mEh MAE, but the correction's accuracy near dissociation rests on an unproven separation of correlation types.

read the letter

The main takeaway is that this paper describes VQE-PT: run standard VQE in a compact active space, extract the 1- and 2-RDMs, and feed them into a classical perturbative correction for the dynamic correlation left out. No bigger circuits or more qubits required. They show this improves energies over plain VQE in simulations on HF and N2, then run F2 on the Quafu processor with error mitigation and report 1.2 millihartree mean absolute error along the full potential energy surface. The RDM-based framing of the correction is the concrete new piece; it keeps the quantum step small while pulling in the missing energy classically. That matches a real need for NISQ chemistry work. The experimental section stands out because it includes hardware data rather than stopping at numerics. The error mitigation appears to have been applied carefully enough to produce usable RDMs. The soft spot is exactly the one flagged in the stress test. The perturbative term assumes the active-space RDMs cleanly encode the dynamic part outside the space. For F2 near dissociation the active-space gap shrinks, static and dynamic correlation mix more, and higher-order perturbative contributions can grow. The simulations on HF and N2 do not test that regime directly, and the abstract gives no detailed comparison to full configuration interaction or larger active spaces to confirm the correction stays accurate. The math itself follows standard perturbation theory on top of VQE, so the citation pattern looks ordinary rather than circular. This is for groups already running VQE on small molecules who want a low-overhead way to push accuracy a bit further on current hardware. A reader focused on experimental quantum chemistry implementations would find the Quafu results worth examining. It deserves peer review because the idea is straightforward, the hardware component is real, and the central assumption can be checked against benchmarks during revision.

Referee Report

2 major / 2 minor

Summary. The paper introduces the perturbative variational quantum eigensolver (VQE-PT), a hybrid quantum-classical algorithm that solves an effective Hamiltonian in a compact active space via VQE and computes a perturbative energy correction from the 1- and 2-RDMs to incorporate dynamic electron correlation outside the active space. Numerical simulations on HF and N2 demonstrate systematic improvements over standard VQE, while an experimental implementation on the Quafu superconducting processor for F2, combined with error mitigation, achieves a mean absolute error of 1.2 millihartree along the potential energy surface.

Significance. If the RDM-based perturbative correction accurately recovers the missing dynamic correlation without substantial corruption from hardware noise, the method offers a resource-efficient route to higher-accuracy VQE chemistry calculations on NISQ hardware by avoiding increases in qubit count or circuit depth. The experimental result on a real superconducting device provides concrete evidence of practicality, though its weight hinges on validation of the static-dynamic partitioning for the chosen systems.

major comments (2)

[Abstract] Abstract: the central experimental claim of 1.2 mEh MAE for the F2 PES on Quafu is load-bearing for the practical-utility conclusion, yet the support rests on the assumption that the second-order correction computed from active-space VQE RDMs faithfully captures dynamic correlation outside the space; no explicit comparison to exact (e.g., FCI or CCSD(T)) dynamic-correlation benchmarks for the same active-space partitioning of F2 is referenced, leaving the assumption unverified especially near dissociation.
[Numerical simulations section] Numerical simulations on HF and N2: while systematic improvements are reported, the manuscript does not quantify the accuracy of the RDM-derived perturbative term against a classical reference that uses the identical active-space RDMs plus exact perturbation; without this, it is unclear whether the observed gains arise from the perturbation itself or from other factors.

minor comments (2)

The abstract refers to 'robust error mitigation strategies' without naming the specific techniques (e.g., zero-noise extrapolation, readout error mitigation) or their parameters; adding a short clause would improve reproducibility.
Notation for the perturbative correction (e.g., the explicit form of the second-order energy expression in terms of the 1- and 2-RDMs) should be stated once in the main text rather than left implicit.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough review and insightful comments on our manuscript introducing VQE-PT. We address each major comment below and have revised the manuscript accordingly to provide additional validation.

read point-by-point responses

Referee: [Abstract] Abstract: the central experimental claim of 1.2 mEh MAE for the F2 PES on Quafu is load-bearing for the practical-utility conclusion, yet the support rests on the assumption that the second-order correction computed from active-space VQE RDMs faithfully captures dynamic correlation outside the space; no explicit comparison to exact (e.g., FCI or CCSD(T)) dynamic-correlation benchmarks for the same active-space partitioning of F2 is referenced, leaving the assumption unverified especially near dissociation.

Authors: We appreciate the referee pointing out the need for explicit validation of the perturbative correction's accuracy. We agree that this strengthens the manuscript. In the revised manuscript, we have added comparisons between the RDM-based second-order correction and exact dynamic correlation energies computed via CCSD(T) for the F2 molecule using the identical active-space partitioning. These comparisons are provided for multiple bond lengths, including near dissociation, demonstrating that the perturbative term accurately recovers the missing correlation with small errors. revision: yes
Referee: [Numerical simulations section] Numerical simulations on HF and N2: while systematic improvements are reported, the manuscript does not quantify the accuracy of the RDM-derived perturbative term against a classical reference that uses the identical active-space RDMs plus exact perturbation; without this, it is unclear whether the observed gains arise from the perturbation itself or from other factors.

Authors: We thank the referee for this observation. To clarify that the improvements stem from the perturbative correction, we have included in the revised numerical simulations section a direct comparison of the RDM-derived perturbative energies to those obtained from a classical calculation using the exact perturbative formula with the same active-space RDMs. This shows close agreement, confirming the validity of our approach and that the gains are attributable to the perturbation. revision: yes

Circularity Check

0 steps flagged

No significant circularity in VQE-PT derivation

full rationale

The paper's core chain is VQE solving the active-space effective Hamiltonian followed by classical computation of a perturbative correction from the resulting 1- and 2-RDMs. This is a standard hybrid partitioning with no self-definitional reduction, no fitted parameter renamed as a prediction, and no load-bearing self-citation or imported uniqueness theorem evident from the abstract or described method. The derivation remains self-contained and independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based solely on the abstract, the method rests on standard assumptions of perturbation theory applied to an effective active-space Hamiltonian; no explicit free parameters or new entities are mentioned.

axioms (1)

domain assumption Perturbation theory applied to the effective Hamiltonian in the active space using reduced density matrices from VQE accurately accounts for dynamic correlation.
Invoked in the description of how the perturbative energy correction is obtained.

pith-pipeline@v0.9.0 · 5755 in / 1148 out tokens · 35067 ms · 2026-05-22T21:27:25.619695+00:00 · methodology

Perturbative Variational Quantum Eigensolver via Reduced Density Matrices

Core claim

What carries the argument

Load-bearing premise

What would settle it

discussion (0)