[Paper Review] The Ising ferromagnet as a self-correcting physical memory: a Monte-Carlo study

Francesca Chadha-Day, S. D. Barrett|arXiv (Cornell University)|Jan 1, 2012

Theoretical and Computational Physics20 references3 citations

TL;DR

This study uses Monte Carlo simulations to investigate the Ising ferromagnet as a classical self-correcting physical memory, showing that in 1D the bit storage lifetime is independent of system size, while in 2D it increases exponentially with system size below the critical temperature $T_c$, confirming the 2D model's superior error resilience due to topological protection from spin-flip fluctuations.

ABSTRACT

The advent of quantum computing has heralded a renewed interest in physical memories - physically realizable structures that offer reliable data storage with error correction only at the point of access. Here, we examine a model of a classical physical memory, capable of storing a classical bit, based on the ferromagnetic Ising model on 1- and 2-dimensional square lattices. We make use of Monte-Carlo simulations as well as analytic solutions in the high temperature limit. At high temperatures, both 1- and 2-D Ising models behave like a non-interacting model, albeit with a reduced effective number of spins and a reduced effective spin-flip rate. In agreement with general arguments, we confirm numerically that (i) the information storage time is independent of system size in 1D, (ii) in two dimensions, at temperatures below the Onsager phase transition temperature T, the information storage time increases exponentially with system size. Interestingly, some benefit in increasing system size is found even above T, although the increase in lifetime is not exponential.

Motivation & Objective

• To investigate whether the classical Ising ferromagnet can function as a self-correcting physical memory by analyzing its bit storage fidelity.
• To compare the performance of 1D and 2D Ising lattices in preserving a stored classical bit under thermal fluctuations.
• To quantify how system size and temperature affect the information storage time using Monte Carlo simulations and effective spin models.
• To explore the emergence of phase transition behavior in the memory’s fidelity scaling with system size.

Proposed method

• Simulate the 1D and 2D Ising model with $J=1$, $h=0$, using Metropolis Monte Carlo dynamics to model thermal relaxation.
• Define the memory fidelity $F(t)$ as the probability of correct bit retrieval via majority vote after time $t$, with error correction only at readout.
• Introduce an effective model with parameters $\lambda$ (bit flip rate) and $N_{\text{eff}}$ (effective number of spins) to describe the decay of fidelity.
• Fit the decay of $F(t)$ to an exponential model $F(t) \sim e^{-\lambda t}$ to extract $\lambda$, and analyze its dependence on system size $N$ and temperature $T$.
• Use high-temperature analytic approximations to validate the effective model and assess its limits.
• Analyze the scaling of $\lambda$ and $N_{\text{eff}}$ with $N$ and $T$ to identify phase transition signatures.

Experimental results

Research questions

• RQ1 Does the 1D Ising model exhibit system-size-independent information storage time, as predicted by theory?
• RQ2 Does the 2D Ising model show an exponential increase in storage time with system size below the critical temperature $T_c$?
• RQ3 How does the fidelity decay rate $\lambda$ scale with system size $N$ in 1D and 2D at temperatures above and below $T_c$?
• RQ4 Can the effective model with $\lambda$ and $N_{\text{eff}}$ quantitatively describe the memory performance across different temperatures and dimensions?
• RQ5 Does the phase transition in the 2D Ising model manifest as a qualitative change in the scaling of $\lambda$ with $N$?

Key findings

• In 1D, the bit flip rate $\lambda$ is independent of system size $N$, confirming that larger systems provide no significant benefit in storage lifetime.
• In 2D, below $T_c$, $\lambda$ decreases exponentially with $N$, indicating that larger systems dramatically extend storage time.
• Above $T_c$, the decay of $\lambda$ with $N$ is sub-exponential, showing a small but non-negligible benefit from increasing system size.
• The effective number of spins $N_{\text{eff}}$ scales linearly with $N$ in 1D, with a prefactor $m \approx 0.61 \pm 0.02$ at $T=3.5$, indicating reduced effective spin count due to interactions.
• In 2D, $N_{\text{eff}}$ shows a negative correlation with $N$ just below $T_c$, reflecting enhanced stability from spin interactions, while it becomes positively correlated above $T_c$, indicating weaker collective protection.
• The phase transition at $T_c \approx 2.269J/k$ is reflected in a qualitative change in the scaling of $\lambda$ with $N$, transitioning from exponential to sub-exponential decay.