Magnetic relaxation of a modified Mattis model

Monte Carlo methods are used to study an Ising model in two dimensions with nearest-neighbor interactions given by $J_{\mathrm{ij}}=\left|J_{\mathrm{ij}}^{\prime }\right|{\xi }_i{\xi }_j$, where ${\xi }_i=\pm 1$ randomly with equal weight and $J_{\mathrm{ij}}^{\prime }$ is a random variable with a Gaussian probability density. Equilibrium properties are obtained both in the paramagnetic phase and in the ordered phase. The model, which is unfrustrated, exhibits approximate logarithmic relaxation. Taken together with other studies, it is concluded that approximate logarithmic relaxation observed in canonical spin glasses does not have its origin in frustration per se, but rather in the continuous distribution of exchange strengths.