The Setting
Consider Glauber dynamics for the critical Sherrington-Kirkpatrick model with zero external field. The starting configuration is weighted by its Gibbs probability, while its distance to equilibrium is measured from that fixed starting state.
Let be the number of sites. A spin configuration is a function on the site set taking values in ; denotes its value at site .✦ An unordered spin pair is an element of , the set of pairs of sites with .✦ A coupling array is a real-valued function on the set of spin pairs.✦ The Gaussian disorder law in (1) has independent coordinates with mean zero and variance : (1)
The Hamiltonian is the function that assigns each configuration its pair interaction energy, as given in (2). The Gibbs distribution is the probability distribution on proportional to ✦; the partition function is the sum of these weights: (2)
Write for the restriction of to all sites other than . Two configurations agree outside when these restrictions coincide.✦ The heat-bath kernel is the transition matrix on that resamples the spin at from its conditional Gibbs distribution, with entries given by (3): (3)✦
The heat-bath generator updates every site at rate one.✦ The continuous-time matrix associated with the heat-bath generator is the matrix exponential of that generator at a real time parameter , and is the transition semigroup for .✦ The random-scan kernel makes one attempt at a uniformly chosen site.✦ With the identity matrix, their formulas are (4): (4)
Unmixed Gibbs Mass
For a real function on , the total variation distance to Gibbs equilibrium is the nonnegative real number in (5): (5)✦
The continuous-time distance , for real , is the total variation distance to equilibrium of the row .✦ The discrete-time distance , for a nonnegative integer , is the distance to equilibrium after attempts from .✦ (6)
The unmixed Gibbs masses and are the Gibbs probabilities of starting configurations whose distances at the indicated time or update count are strictly greater than . (7)
The worst-start mixing times and are the infima of the nonnegative times and update counts at which every starting state is within of equilibrium. (8)
The Claim
Theorem 1. Let be deterministic times with , and let be deterministic update counts with . The equilibrium-start assertion is (9), where convergence is in probability under the disorder laws : (9)✦ Let . (10)✦ (11)✦ The worst-start mixing bounds are (12) and (13): (12)✦ (13)✦ For every fixed , the power bounds (14) hold: (14)✦ All limits are taken as .
What Is Checked
The statements concern the Gibbs model (2) under the Gaussian disorder law (1), with the heat-bath dynamics (4) and deterministic nonnegative time or update sequences on the stated scales.