Classical Optimization Strategies for ADAPT-VQE for Chemical Applications

The ADAPT-VQE approach systematically generates a compact ansatz based on derivatives of the electronic energy function, and previous work has demonstrated its success for molecular systems with small basis sets. Here, we describe the implementation of ADAPT-VQE with our recently developed sparse wavefunction circuit solver (SWCS) in terms of accuracy and efficiency for molecular systems with up to 52 spin-orbitals. The SWCS is easily tuned with one parameter to balance computational cost and accuracy, which extends the application of ADAPT-VQE for molecular electronic structure calculations to larger basis sets. Alternative optimization strategies for ADAPT-VQE are discussed that reduce the computational cost of the optimization. By pre-optimization of the wavefunction ansatz via ADAPT-VQE/SWCS on classical hardware, we aim to utilize the power of classical high-performance computing to reduce the work required on noisy intermediate-scale quantum hardware.