Variational Quantum Eigensolver (VQE) Simulations

• Variational Quantum Eigensolver (VQE) is a quantum-classical hybrid algorithm that approximates ground-state energies of many-body systems using parameterized quantum circuits and classical optimization.
• It employs fermion-to-qubit mappings such as Jordan–Wigner and advanced ansätze like UCCSD to convert electronic-structure Hamiltonians for execution on NISQ hardware.
• Recent advancements focus on noise mitigation, adaptive ansatz strategies, and efficient measurement grouping to scale VQE for complex molecular and condensed matter simulations.

The Variational Quantum Eigensolver (VQE) is a quantum-classical hybrid algorithm designed to approximate ground-state properties of quantum many-body systems, particularly the electronic structure of atoms and molecules. VQE leverages a parameterized quantum circuit and a classical optimization loop to minimize the expectation value of a mapped qubit Hamiltonian, exploiting the Rayleigh–Ritz variational principle. With the current generation of noisy intermediate-scale quantum (NISQ) devices, VQE has emerged as the most widely implemented algorithm for quantum chemistry and many-body physics, constituting a reference framework for both methodological development and benchmarking across platforms.

1. Electronic-Structure Hamiltonians and Qubit Mapping

In VQE simulations for molecular systems, the computation begins from the second-quantized electronic Hamiltonian

$$\hat{H}_e = \sum_{pq} h_{pq} \hat{a}_p^\dagger \hat{a}_q + \frac{1}{2} \sum_{pqrs} g_{pqrs} \hat{a}_p^\dagger \hat{a}_q^\dagger \hat{a}_r \hat{a}_s$$

where $h_{pq}$, $g_{pqrs}$ denote one- and two-electron integrals, typically obtained from a classical quantum chemistry package (e.g., PySCF in the STO-3G basis) (Anaya et al., 2022).

To deploy this on a quantum processor, the fermionic operators are mapped to qubit operators via Jordan–Wigner, Parity, or Bravyi-Kitaev schemes. This yields a qubit Hamiltonian,

$$H = \sum_{i} c_i P_i, \qquad P_i \in \{I, X, Y, Z\}^{\otimes n}$$

where each $P_i$ acts on $n$ qubits, and the $c_i$ coefficients are determined by the integrals and chosen mapping.

For molecules with symmetries, further qubit count reduction is possible; e.g., the two-qubit reduction exploits $\mathbb{Z}_2$ parity (Anaya et al., 2022).

2. Variational Principle and Ansatz Construction

VQE implements the Rayleigh-Ritz variational principle by preparing a parametric quantum state $|\psi(\theta)\rangle = U(\theta) |\Phi_0\rangle$ via a gate-based circuit $U(\theta)$. The goal is to minimize the energy functional

$$E(\theta) = \langle \psi(\theta) | H | \psi(\theta) \rangle \geq E_0$$

where $E_0$ is the true ground-state energy (Anaya et al., 2022). The minimization is performed over the parameter vector $\theta$.

Ansatz construction in chemistry applications often employs the Unitary Coupled Cluster Singles and Doubles (UCCSD) form, which can be written as

$$U(\theta) = \exp\left[ \sum_{k} \theta_k (T_k - T_k^\dagger) \right]$$

where $T$ are excitation operators. Gate-by-gate details are generated via dedicated libraries, e.g., Qiskit Chemistry, which implements UCCSD using Hartree–Fock reference states, Givens rotations, and controlled entanglers (Anaya et al., 2022).

For many-body and condensed matter models such as the Heisenberg chain, QAOA-inspired alternating-layer ansätze using exponentials of commuting Hamiltonian halves are efficacious (Wang et al., 2023).

3. Measurement, Grouping, and Error Analysis

The Hamiltonian $H$ is decomposed into a sum of Pauli strings, each requiring separate expectation value measurement. Qiskit and related frameworks automatically group commuting Pauli terms to minimize the number of required circuit executions per optimization step. For shot-based (QASM) simulations, the user specifies the number of measurements ('shots') to achieve statistical accuracy; e.g., 1024 per group for H$_2$ ground-state computations (Anaya et al., 2022).

Precision assessments show that, for small molecules or atoms, state-vector simulations consistently reach chemical accuracy, while QASM-based methods require $\mathcal{O}(10^4-10^5)$ shots per measurement to suppress sampling error to sub-milliHartree levels (Sumeet et al., 2021). For example, Be atom ground-state simulations in STO-3G, with 20,000 shots, converged to $\sim$0.1% of the FCI benchmark energy, and errors from shot noise were comparable to the gap between classical CCSD and VQE-UCCSD estimates (Sumeet et al., 2021).

4. Classical Optimization Loops and Resource Scaling

Classical optimizers (SLSQP, COBYLA, SPSA) iteratively update $\theta$ to minimize $E(\theta)$. The workflow consists of repeated preparation, measurement, expectation construction, and parameter update stages. Convergence is typically declared when energy gradients fall below a set threshold or maximum iteration counts are reached (Anaya et al., 2022, Ion et al., 27 Dec 2025).

Resource scaling is dictated by qubit counts (twice the number of spin orbitals before tapering), circuit depth (driven by ansatz layer count and entangler complexity), and measurement overhead (number of Pauli terms $\times$ shots per term). For UCCSD, the parameter and gate count scales as $O(N^4)$ (with $N$ the number of orbitals), and measurement cost as $O(N^7/\epsilon^2)$ for energy accuracy $\epsilon$ (Singh et al., 2024). Parity mappings and two-qubit reductions are crucial for halving hardware requirements.

Adaptive ansätze and problem-inspired layerings (QAOA-, Hamiltonian-, and singular-value-based) are imperative for scaling VQE to larger qubit registers and deeper correlation regimes (Wang et al., 2023, Yordanov et al., 2020).

5. Benchmarks, Accuracy Comparisons, and Physical Observables

Simulations of the electronic ground state of H$_2$ (bond length $0.74$ Å) show VQE convergence within $10^{-4}$ Hartree of exact diagonalization in 50–100 steps (Anaya et al., 2022). Extensive benchmarks on Be, Li$^{-}$, B$^{+}$, and H$_2$O confirm that VQE-UCCSD recovers nearly all electron correlation within a basis—comparable to FCI and traditional CCSD—in both statevector and QASM modes, with QASM energies retaining $\sim$90 – 99% of FCI correlation for sufficiently large shot counts (Sumeet et al., 2021, Ion et al., 27 Dec 2025).

Physical observables such as energy, molecular geometry, spin, and electron number are accessible within the VQE framework, with specialized approaches for force calculations (Hellmann-Feynman theorem, finite-difference central-difference schemes) in molecular geometry optimization tasks (R et al., 2022). For periodic and strongly correlated systems, tailored orbital rotations and subspace expansions (K2G, VQE/QSE) remedy accuracy deficits and enable excited-state calculations (Liu et al., 2020).

6. Scalability, Advanced Algorithms, and Quantum Hardware Implementation

Experiments demonstrate that VQE can scale to complex systems with up to 10–20 qubits (organic molecules, atomic ions), subject to ansatz expressivity and shot-budget constraints. Variational approaches for Heisenberg and XXZ chains up to $N=28$ qubits achieve sub-milliJ accuracy, with measurement sampling overhead scaling polynomially in $N$ (Wang et al., 2023). Adaptive ansätze, QAOA-type layerings, and efficient initializations significantly reduce circuit depth and improve convergence (Yordanov et al., 2020).

On current hardware, exact wavefunction simulation is limited to $N\lesssim 20$; sampling-based VQE with efficient ansatz and measurement strategies extends practical system size, and resource scaling favors problem-adapted layerings over generic UCCSD (Singh et al., 2024, Anaya et al., 2022).

Recent studies emphasize the critical need for measurement cost reduction and noise resilience. Partial Hamiltonian methods (SHARC-VQE) reduce measurement overhead from $O(N^4/\epsilon^2)$ to $O(1/\epsilon^2)$ per energy estimate, and total VQE cost from $O(N^7/\epsilon^2)$ to $O(N^3/\epsilon^2)$, substantially suppressing noise-induced errors to 5–10% without explicit mitigation (Singh et al., 2024).

7. Extensions and Outlook

VQE serves as a foundation for hybrid quantum algorithms and precision quantum simulations in atomic, molecular, nuclear, and condensed matter physics. Its methodological flexibility enables integration with classical post-processing (neural hybrid eigensolver, adiabatic connection corrections), periodic boundary conditions, and adaptive parameterization, facilitating robust ground- and excited-state calculations beyond the reach of classical simulation.

Ongoing developments address challenges of circuit depth, optimizer efficiency, shot-budget allocation, and scalability to larger qubit registers in anticipation of fault-tolerant quantum devices. The VQE paradigm remains central to exploring quantum advantage across molecular and many-body simulation domains (Anaya et al., 2022, Sumeet et al., 2021, Wang et al., 2023, Singh et al., 2024).