VQE for Quantum Chemistry
Using variational eigensolvers to estimate molecular ground-state energies, the flagship near-term application of quantum simulation.
The chemistry problem
The central task of quantum chemistry is solving the electronic Schrodinger equation: find the ground-state energy of electrons moving in the field of fixed nuclei. Because electrons are strongly correlated fermions, the classical cost of exact solutions grows factorially with system size, which is why chemistry is a leading target for quantum methods.
From molecule to qubits
- Choose a basis set and compute one- and two-electron integrals classically.
- Write the second-quantized Hamiltonian in terms of fermionic creation and annihilation operators.
- Map fermions to qubits via Jordan-Wigner, Bravyi-Kitaev, or parity encoding.
- The result is a qubit Hamiltonian H = sum_l c_l P_l over Pauli strings.
Chemical accuracy
The community benchmark is chemical accuracy, roughly 1.6 milli-Hartree (about 1 kcal/mol), the precision needed to predict reaction rates reliably. VQE aims to reach this for molecules where classical coupled cluster or full configuration interaction becomes expensive.
Demonstrations and their scale
Early experiments computed the bond-dissociation curve of H2, then LiH, BeH2, and small hydrogen chains on superconducting and trapped-ion hardware. These are within reach of classical exact diagonalization; they validate the workflow rather than achieve advantage. Scaling to strongly correlated molecules beyond classical reach remains the open goal.
The obstacles
Three bottlenecks dominate: measurement overhead (many Pauli terms, each needing many shots), optimization difficulty (barren plateaus and noisy gradients), and hardware noise corrupting deep ansatze. Active-space reduction, term grouping, error mitigation, and compact ansatze such as ADAPT-VQE all address these. Whether VQE reaches practical advantage before fault-tolerant phase estimation does is an open question; many expect the decisive chemistry results to come from qubitization-based phase estimation on error-corrected machines.