2025 Energy HPC Conference

Hybrid Quantum/Classical Machine Learning for Molecular Conformation Generation

Abstract: We present an algorithm for hybrid quantum/classical computing environments that generates low-energy conformations of small and medium size hydrocarbon molecules. Despite the importance of conformers in determining physical and chemical properties, traditional physical solvers often struggle to find low-energy conformers due to the large search space. To address this issue, we investigate the potential of using a hybrid generative adversarial network (GAN) algorithm. This algorithm trains a hybrid quantum/classical generator using a simulated photonic quantum processor and a GPU on a dataset of alkane molecules to generate conformers with a specified energy. We find the use of a quantum processor leads to higher-quality results, with the hybrid GAN producing conformers up to 50% closer to the target energy than an equivalent classical GAN.

Authors: William Clements (ORCA Computing), Hugo Wallner (ORCA Computing), Corneliu Buda (bp),
Omar Bacarreza (ORCA Computing), Peter Lemke (bp) and Claudia Perry (bp)