arXiv:2609.33975
Swapping Quantum Annealing Errors into a Cavity
Hao Zhang
Abstract
Quantum annealing is a promising form of quantum computation, but it slows down at a small energy gap $Δ$ during the anneal, and the errors it makes there remain undetected. We introduce cavity quantum annealing, in which a cavity mode of frequency $O(1)\ggΔ$, coupled mid-schedule, swaps these errors into photons. In the $p$-spin model, this reduces the annealing time to nearly the square root of that required without the cavity. Moreover, counting the photons certifies the result: high-photon runs, with an $O(1)$ yield, reach a ground-state fidelity above $99.99\%$ orders of magnitude sooner, a gain that grows exponentially with system size. A pre-loaded photon instead runs the swap in reverse, preparing an excited state. More broadly, our work shows that cavities can enhance computations performed with qubits, paving the way toward high-fidelity quantum state preparation by quantum annealing.
AI Summary
The paper asks whether excitations produced at a small annealing gap can be transferred to an auxiliary cavity, both improving ground-state preparation and providing a measurement record of the transfer. It studies a coherently evolving ferromagnetic $p=3$ collective-spin model coupled to one or two cavity modes through a mid-schedule pulse, with coupling vanishing at both endpoints. Exact Schrödinger evolution in the permutation-symmetric spin sector shows that, after the first-order transition, cavity-induced avoided crossings transfer population from excited spin levels to the spin ground state plus photons. A Landau–Zener cascade model explains the error suppression through these crossings. Over the simulated sizes, the time to reach spin error $0.1$ fits $T\propto e^{0.087N}$ with one mode, versus $e^{0.16N}$ without the cavity; two modes give a fitted exponent $0.080$. A semiclassical ladder-overlap calculation reproduces these exponents using a numerically determined cutoff growth rate. Photon counting provides a conditional ground-state indicator rather than an unconditional certificate. At $N=24$, three-photon outcomes have fidelity above $0.97$ from $T=12$, but their average yield over $T>10$ is about $0.1\%$. Scaling comparisons report conditional errors down to $10^{-4}$ while requiring yield at least $10^{-4}$. Preloading one photon reverses the transfer and prepares a spin excitation. The demonstrated speedup remains exponential and concerns this collective model. Whether additional modes can change the scaling qualitatively remains open; applying the mechanism to other models requires suitable coupling schemes, and experimental tests remain future work.
Central claim may not hold: The near-square-root speedup is supported by finite-size fits and a cutoff model using a dynamically measured growth rate, rather than an independently established asymptotic scaling law. The abstract's O(1) certification yield is not established asymptotically: the reported yields and the imposed 10^−4 acceptance floor support only the studied finite-size regime.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 10 pages, 6 figures