Critical scaling advantage in spin glass quantum optimization

Synopsis

More than two decades ago, experiments on disordered alloys suggested that spin-glasses could be brought into low-energy states faster through quantum annealing than through conventional thermal annealing. Replicating this phenomenon in a programmable processor has remained a central challenge in quantum optimization. In this presentation, I will provide the first experimental demonstration of the critical scaling advantage of quantum annealing compared to classical algorithms, utilizing a superconducting quantum annealing processor with thousands of qubits [1]. I will offer a theoretical explanation of this advantage through the dynamics of quantum phase transition in spin-glasses. By extracting critical exponents, I will show a clear distinction between quantum annealing and the slower stochastic dynamics of analogous Monte Carlo algorithms. Finally, I will present recent results on Quantum Error Mitigation as a near-term approach to reduce errors when estimating expectation values [2]. Using Zero-Noise Extrapolation in the quantum critical dynamics of a transverse-field Ising spin-chain, I will present the successful mitigation of static control errors and thermal noise, extending coherent annealing time by nearly an order of magnitude.

[1] King et al., Nature 617, 61 (2023).
[2] Amin et. al. arXiv:2311.01306.