Optical spin glasses: a new model for glassy systems

TL;DR

This paper proposes an optical system with adaptive mirrors that mimics spin glass Hamiltonians, enabling large-scale simulations with linear scaling, potentially surpassing traditional computer simulations in efficiency.

Contribution

It introduces a novel optical setup for simulating spin glasses, achieving linear scaling in simulation steps, unlike the quadratic or cubic scaling in conventional methods.

Findings

Optical system can replicate spin glass Hamiltonian mathematically.

Simulation step complexity scales linearly with number of spins.

Potential for more efficient large-scale spin glass simulations.

Abstract

The aim of this work is to prove that it is possible to realise an optical system which produces as output a light intensity that can be expressed in the same mathematical form of the spin glass Hamiltonian. The optical system under study is controlled through an adaptive optics device composed by millions of switchable mirrors in an ON and OFF position. These user controlled mirrors are playing the role of the spin states. Furthermore, such optical system can be used to run simulations and, applying the same Metropolis Monte Carlo algorithm used in computer simulations, extract the quantities of interest for Spin Glass physics. The proposed system has a great advantage over existing computer-based spin glass simulations: the simulation step has no scaling dependence on the total number of spins $N$ and the whole simulation has a linear dependence on $N$, due to the fact that we want to keep the number of moves per spin constant. This should be compared to the $N^2$ dependence of a single Monte Carlo move in computer simulation, which yields a global N3 dependence. Some experimental problems limit such advantage, but are partially addressed at the end of the presentation.