SciPost Submission Page
Accelerating equilibrium spin-glass simulations using quantum annealers via generative deep learning
by Giuseppe Scriva, Emanuele Costa, Benjamin McNaughton, Sebastiano Pilati
This Submission thread is now published as
Submission summary
| Authors (as registered SciPost users): | Sebastiano Pilati · Giuseppe Scriva |
| Submission information | |
|---|---|
| Preprint Link: | https://arxiv.org/abs/2210.11288v2 (pdf) |
| Code repository: | https://doi.org/10.5281/zenodo.7118502 |
| Data repository: | https://doi.org/10.5281/zenodo.7250436 |
| Date accepted: | 2023-05-22 |
| Date submitted: | 2023-04-28 22:46 |
| Submitted by: | Scriva, Giuseppe |
| Submitted to: | SciPost Physics |
| Ontological classification | |
|---|---|
| Academic field: | Physics |
| Specialties: |
|
| Approach: | Computational |
Abstract
Adiabatic quantum computers, such as the quantum annealers commercialized by D-Wave Systems Inc., are routinely used to tackle combinatorial optimization problems. In this article, we show how to exploit them to accelerate equilibrium Markov chain Monte Carlo simulations of computationally challenging spin-glass models at low but finite temperatures. This is achieved by training generative neural networks on data produced by a D-Wave quantum annealer, and then using them to generate smart proposals for the Metropolis-Hastings algorithm. In particular, we explore hybrid schemes by combining single spin-flip and neural proposals, as well as D-Wave and classical Monte Carlo training data. The hybrid algorithm outperforms the single spin-flip Metropolis-Hastings algorithm. It is competitive with parallel tempering in terms of correlation times, with the significant benefit of a much shorter equilibration time.
Author comments upon resubmission
we thank the two Referees for their positive assessments of our manuscript. Both recognize that it presents a promising application of quantum annealers to statistical mechanics problems, and they recommend publication in SciPost Physics. They also request us to provide a few clarifications. We do so in the revised manuscript. Chiefly:
i. As suggested by Referee 1, we clearly describe the role of the annealing time on the sampled energies and the distinction from the physical temperature of the devi e;
ii. As suggested by Referee 2, we provide a short report on recent research on improved MC sampling algorithms, and we mention the possible connection with our work.
We hope that with these clarifications our manuscript will be processed further towards publication.
Sincerely,
the authors
List of changes
i. In the "Conclusions" section, an extended discussion on the role of the annealing times and on the physical temperature of the device is included.
ii. In the "Conclusions" section, a short report on recent research on improved MC algorithms is provided.
iii. The additional references from [70] to [80] have been included.
Published as SciPost Phys. 15, 018 (2023)