A hybrid system that processes data through a quantum computer before training conventional AI delivered about 20 percent greater accuracy in predicting chaotic physical systems, while requiring hundreds of times less memory than standard models that rely on classical computing alone.
The method, developed by researchers at University College London and published in Science Advances, addresses a fundamental tension in scientific computing. Full simulations of complex systems like fluid flow can take weeks, often too long to be useful. AI models offer speed but become unreliable over longer time scales. The quantum-informed approach attempts to capture the best of both: rapid predictions that remain stable.
Professor Peter Coveney, senior author from UCL Chemistry and the Advanced Research Computing Centre, described the advance:
"To make predictions about complex systems, we can either run a full simulation, which might take weeks -- often too long to be useful -- 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. Making predictions about fluid flow and turbulence is a fundamental science challenge but it also has many applications. Our method can be used in climate forecasting, in modeling blood flow and the interaction of molecules, or to better design wind farms so they generate more energy."
The workflow uses the quantum computer at a single point: identifying invariant statistical properties, patterns that remain stable over time, within training data generated by simulations or observations. These quantum-derived patterns then guide the training of an AI model running on a conventional supercomputer.
This limited quantum role appears deliberate. Current quantum hardware faces persistent challenges from noise, errors, and interference, which typically demand large numbers of measurements. By using the quantum computer only once rather than repeatedly exchanging data between quantum and classical systems, the method sidesteps these limitations.
The performance gains stem from how quantum computers represent information. Traditional computers use bits set to either 1 or 0. Quantum computers use qubits, which through superposition can exist as 1, 0, or anything in between until measured.
Through entanglement, each qubit can influence others regardless of distance. Together, these properties allow a relatively small number of qubits to represent an enormous number of possible states, compacting information in ways classical systems cannot match.
The researchers suggest this compression mirrors something about the physical systems themselves. Many complex systems behave in ways that resemble quantum effects, where changes in one region influence distant parts, similar to entanglement. The quantum computer, they propose, is particularly well suited to capturing this underlying physics.
First author Maida Wang of the UCL Centre for Computational Science characterized the achievement:
"Our new method appears to demonstrate 'quantum advantage' in a practical way – that is, the quantum computer outperforms what is possible through classical computing alone. These findings could inspire the development of novel classical approaches that achieve even higher accuracy, though they would likely lack the remarkable data compression and parameter efficiency offered by our method. The next steps are to scale up the method using larger datasets and to apply it to real-world situations which typically involve even more complexity. In addition, a provable theoretical framework will be proposed."
The study used a 20-qubit IQM quantum computer connected to classical computing resources at the Leibniz Supercomputing Centre in Germany. Quantum computers must operate at temperatures around minus 273C, close to absolute zero.
Co-first author Xiao Xue, from Advanced Research Computing at UCL, noted:
"In this work, we demonstrate for the first time that quantum computing can be meaningfully integrated with classical machine learning methods to tackle complex dynamical systems, including fluid mechanics. It is exciting to see this kind of 'quantum-informed' approach moving towards practical use."
The research was funded by UCL and the UK's Engineering and Physical Sciences Research Council, with additional support from IQM Quantum Computers and the Leibniz Supercomputing Centre.
Science Advances, 12(16). https://doi.org/10.1126/sciadv.aec5049