Researchers Develop New Quantum Ansatz for Molecular Energy Calculations

A new method for calculating molecular ground-state energies improves computational drug discovery and understanding of chemical reactions. Azadeh Alavi and colleagues at RMIT University, in collaboration with Pattern Recognition Pty Ltd, present a fixed-topology Givens-exchange ansatz that achieves chemically accurate results for lithium hydride, water, and beryllium hydride without complex architecture searches. Their approach, detailed through six independent runs, demonstrates mean errors as low as 0.000000124 Hartree, lower than those achieved by methods using quantum architecture search, providing a reproducible and accurate reference template for variational quantum eigensolvers.

Fixed-topology ansatz delivers unprecedented accuracy in molecular ground-state energy calculations

Errors in calculating molecular ground-state energies dropped to 0.000000124 Hartree for lithium hydride, a level of precision previously unattainable with quantum architecture search methods. This new level of accuracy establishes a benchmark for simulating molecular behaviour, as existing methods struggled to reach such detail without extensive circuit design optimisation. By employing a fixed-topology Givens-exchange ansatz, complex searches were circumvented, delivering a reproducible and reliable tool for variational quantum eigensolvers; this standardised approach promises to accelerate progress in computational drug discovery and materials science.

Multiple molecular benchmarks consistently demonstrated this accuracy, achieving a mean error of 0.000128558 Hartree, equivalent to 0.128558 milli-Hartree, for water molecules using an eight-qubit system. Calculations for beryllium hydride yielded an even lower average error of 0.000002152 Hartree, or 0.002152 milli-Hartree. The fixed-topology Givens-exchange ansatz, a standardised approach to building quantum circuits, also proved chemically accurate in every run for lithium hydride-6, utilising a six-seed evaluation protocol to ensure reproducibility and reliability.

The method leverages the Stiefel manifold, a mathematical framework for unitary evolution, and Givens exchange macros efficiently redistribute electron behaviour within the molecular model; this contrasts with reinforcement learning methods that require iterative circuit construction and parameter adjustments. Although the fixed-topology Givens-exchange ansatz delivers chemically accurate results for small molecules without searching for optimal quantum circuit designs, its current demonstration remains limited. The demonstrations utilising lithium hydride, water, and beryllium hydride represent an initial, limited scope, but the method’s strength lies in providing a dependable, pre-defined structure for calculating molecular energies, bypassing the computationally intensive process of designing custom quantum circuits.

Consequently, it establishes a valuable benchmark for future development of variational quantum eigensolvers, even with its present constraints on molecular complexity and scale. The team at RMIT University, CSIRO, and Pattern Recognition Pty Ltd has established a new, dependable method for calculating the ground-state energies of molecules, a vital step in designing new materials and pharmaceuticals. Their approach, utilising a fixed-topology Givens-exchange ansatz, a standardised ‘recipe’ for building quantum circuits, avoids time-consuming searches for optimal circuit designs, offering a reproducible reference template for future development of variational quantum eigensolvers.

The researchers demonstrated a method for calculating molecular ground-state energies with high accuracy for lithium hydride, water, and beryllium hydride. Achieving errors of up to 0.128558 milli-Hartree for water molecules, this standardised approach to quantum circuit design avoids the need for complex architecture searches. This provides a dependable and reproducible method for determining these energies, which are important in computational drug discovery and materials science. The team intends for this work to serve as a benchmark for further development of variational quantum eigensolvers.