Quantum Computing Enabled ab initio Molecular Dynamics Simulations

As quantum computing matures, a key challenge is integrating quantum processors into practical chemical workflows. This study investigates whether current quantum hardware can serve as the electronic-structure engine for ab initio molecular dynamics (AIMD), a technique typically reserved for classical high-performance computing. Specifically, the authors test if a quantum-classical hybrid approach can accurately propagate nuclear trajectories in both gas-phase and complex, condensed-phase (solvated) environments.

The authors utilize a workflow coupling the Amber molecular dynamics engine with a quantum-centric solver. The core of this approach is Sample-based Quantum Diagonalization (SQD), which uses a quantum processor to sample important electronic configurations (determinants) from a chemistry-inspired LUCJ ansatz. These samples are then used to construct a compact Hamiltonian subspace, which is diagonalized classically to recover accurate energies and analytical nuclear gradients. The team benchmarked this method against exact Full Configuration Interaction (FCI) calculations in the STO-3G basis for ammonia, methane, and water, both in isolation and embedded in explicit liquid water.

The SQD-driven AIMD simulations successfully reproduced FCI-level energy fluctuations and RMS gradient profiles. In gas-phase tests, the method maintained energy agreement within 1 kcal mol-1 of the FCI reference. In condensed-phase QM/MM simulations, the quantum-driven trajectories accurately captured solute-solvent structural properties, such as radial distribution functions, demonstrating that the quantum engine can provide stable forces for complex environments. The authors highlight that SQD effectively compresses the Hilbert space, allowing for accurate dynamics by diagonalizing in a small fraction of the total determinant space.