GPU-accelerated calculations for carbon capture materials on simulated QPUs: scaling to larger active spaces

Quantum computers have the potential to improve the accuracy of material calculations at the quantum scale, enabling unprecedented computational studies of material physics and chemistry. However, near-term machines are limited both by number of qubits and their high error rate. Many contemporary methods leverage embedding theories to study active regions of a molecular system, while treating the surrounding environment in a mean-field manner. Even still, the active spaces of systems of practical interest require qubit counts that are larger than available quantum hardware. Simulating noiseless qubits on GPUs allows one to study material systems with larger active spaces than is currently possible with existing hardware. In this talk, we will discuss our work studying molecular analogs of amine-functionalized metal-organic frameworks, a promising class of materials for the capture of CO₂ and other molecular species from air or other sources.

In particular, we will discuss algorithmic advances, such as gate fusion, that have allowed us to simulate systems with up to 24 qubits with a variational quantum eigensolver. We will additionally explore the effects of the classical components of the algorithm, such as the optimizer, initial state guess, etc. and how they relate to the energy convergence of the systems studied.