Quantum computers improve AI predictions

An AI model trained on calculations from a quantum computer can predict the behaviour of a complex physical system over long periods more accurately than today’s best models, which rely solely on information processed by classical computers. This is shown by a study led by University College London (UCL). The results, published in the scientific journal Science Advances, could be used to optimise models of fluid dynamics that depict the motion and interaction of liquids and gases and are applied in climate, transport and medical research.

For the study, UCL researchers used a 20‑qubit quantum computer from IQM Quantum Computers as well as supercomputing resources at the Leibniz Supercomputing Centre (LRZ). They attribute their findings to the ability of quantum systems to process large volumes of data more efficiently. Unlike classical computers, the computing units of a quantum computer — qubits — do not operate solely with 0 and 1, but with additional states. When qubits are entangled, their computational power increases further. As a result, quantum computers can calculate or analyse data simultaneously and can handle computations that push classical computers to their limits.

“To make predictions about complex systems, we can either run a full simulation, which might take weeks – often too long to be useful” says Professor Peter Coveney, lead author of the study and Professor of Chemistry and of the Advanced Research Computing Centre at UCL. “Or we can use an AI model which is quicker but more unreliable over longer time scales. Our quantum-informed AI model means we could provide more accurate predictions quickly.”

Background:

• You can find the paper featuring insights into technologies used at Science Advances.
• UCL reports in their news section.