Variational Quantum Eigensolver Workflow

The hybrid loop that estimates a Hamiltonian's ground-state energy by minimizing a parameterized quantum circuit's expectation value.

The variational principle

For any Hamiltonian H and any normalized state |psi>, the expectation is at least the ground-state energy E_0. The Variational Quantum Eigensolver (VQE) prepares a parameterized trial state |psi(theta)> with a quantum circuit and uses a classical optimizer to lower toward E_0. The quantum device supplies the hard-to-compute expectation; the classical computer steers the parameters.

The loop

• Choose an ansatz circuit U(theta) and prepare |psi(theta)> = U(theta)|0>.
• Decompose H into a weighted sum of Pauli strings H = sum_k c_k P_k.
• Measure each in the appropriate basis and combine as sum_k c_k .
• A classical optimizer updates theta to reduce the energy.
• Repeat to convergence.

Measuring the Hamiltonian

Because a quantum measurement yields Pauli expectation values, H must be expressed in Pauli terms. For molecular problems this decomposition comes from a fermion-to-qubit mapping such as Jordan-Wigner or Bravyi-Kitaev. Terms that commute can be measured together, reducing the number of distinct measurement circuits; grouping strategies substantially cut shot cost.

Ansatz choice

The ansatz trades expressibility against trainability. Chemistry-motivated choices like UCCSD respect physical structure but are deep; hardware-efficient ansaetze are shallow but risk barren plateaus and may miss the true ground state. The right balance depends on the problem and the device.

Strengths and caveats

VQE keeps circuits shallow relative to phase estimation, making it a leading near-term approach. Its accuracy is limited by ansatz expressibility, measurement noise, and optimizer performance. It gives an upper bound on E_0, not a certified value; excited states require extensions such as subspace or folded-spectrum methods.