An approach to chaos in some mixed \(p\)-spin models (Q377505)

The authors develop an approach to chaos in disorder and temperature for mixed \(p\)-spin models which is based on a novel application of the \textit{S. Ghirlanda} and \textit{F. Guerra} identities [J. Phys. A, Math. Gen. 31, No. 46, 9149--9155 (1998; Zbl 0953.82037)], used here to derive a new family of identities in the setting of the coupled systems. To this end, the authors consider pure \(p\)-spin Hamiltonians \(H_{N,p}(\sigma)\) for \(p\geq 1\) indexed by \(\sigma\in\Sigma_N=\)\{\(-1,+1\)\}\(^N\), \(N\geq 1\), \[ H_{N,p}(\sigma)=\frac{1}{N^{(p-1)/2}}\sum_{1\leq i_i\dots i_p\leq N}g_{i_1\dots i_p}\sigma_{i_1}\dots\sigma_{i_p}, \] where random variables \((g_{i_1\dots i_p})\) are standard Gaussian independent for all \((i_1\dots i_p)\) and \(p\geq 1\). The covariance of this Gaussian process is computed and is given by \[ \frac{1}{N}{E}H_{N,p}(\sigma^1)H_{N,p}(\sigma^2) =(R(\sigma^1,\sigma^2))^p, \] where quantity \(R(\sigma^1,\sigma^2) =N^{-1}\sum_{i=1}^N\sigma_i^1\sigma_i^2\) is the overlap of spin configurations \(\sigma^1,\sigma^2\in\Sigma_N\). A mixed \(p\)-spin Hamiltonian is defined by a linear combination \[ H_{N}(\sigma)=\sum_{p\geq 1}\beta_pH_{N,p}\sigma \] with coefficients \((\beta_p)\) that decrease fast enough to ensure that the process is well defined, \(\sum_{p\geq 1}2^p\beta_p^2<\infty.\) The Gibbs measure \(G_N(\sigma )\) on \(\Sigma_N\) is defined by \[ G_N(\sigma)=\frac{\exp H_{N}(\sigma)}{Z_N}, \] where \({Z_N}\) is the normalizing factor.