Quantum annealing processors achieve computational advantage in simulating problems on quantum entanglement

(Press-News.org) Quantum annealing processors outperform classical supercomputers in solving real-world scientific simulations of quantum spin dynamics, researchers report in a new study, achieving results far beyond the capacity of conventional computational methods, which may require impossible time and energy to match. The results provide a challenge to classical computing, where method improvement has in the past tempered claims of quantum advantage. Only in recent years have quantum computers begun to live up to their lofty promises, with quantum processing units (QPUs) with diverse architectures – such as photonic, neutral-atom, and superconducting systems – beginning to surpass even the most powerful supercomputers in solving complex problems. However, while it is now widely accepted that current QPU technologies outperform classical methods in certain tasks, such as random-number generation, hardware imperfections have limited the advantage of quantum processors over classical computation in practical scientific applications, making it difficult to demonstrate clear cases of quantum superiority. Here, Andrew King and colleagues evaluate the performance of superconducting quantum annealing (QA) processors in simulating a more complex problem – the continuous-time quantum dynamics of the transverse-field Ising model (TFIM). To benchmark the QPUs performance, King et al. compared the results to high-precision matrix product state (MPS) simulations run on powerful classical supercomputers and used advanced classical techniques, such as tensor networks and neural networks, to estimate the cost of approximating quantum dynamics to match QPU accuracy. According to the findings, the quantum processor outperformed classical MPS simulations across a range of Ising model topologies. Moreover, resource requirement estimates for classical simulation reveal severe limitations. To match QPU performance, the authors estimate that MPS methods would require computational resources far beyond practical feasibility, including millions of years of supercomputing time and electricity requirements that exceed annual global consumption. “This impracticability of classical simulation opens the door to quantum advantage in optimization and AI, addressing scientific questions that may otherwise remain unanswered, and applications that may be classically impossible,” write King et al.

END