The Setting
Equilibrium initialization asks about typical starting configurations under the Gibbs law. Linear combinations of the spins give a family of observables for measuring equilibrium fluctuations and their Dirichlet cost.
Let be a nonnegative integer, with sites indexed by . A spin configuration assigns a spin value to each site .✦ A coupling array assigns a real number to each unordered pair of distinct sites, represented by .✦ The Hamiltonian of a coupling array on sites is the real function on giving the pair interaction energy in (1). (1)✦
For a coupling array on sites, the partition function is the sum of the weights over , and the Gibbs distribution is the probability distribution on obtained by dividing these weights by , as in (2). (2)
The Gibbs mean of a real observable on is its average under ; its Gibbs variance is its mean squared deviation from that average, as in (3). (3)
For a site and a configuration , let be the set of configurations agreeing with outside .✦ The heat-bath kernel replaces the spin at according to the Gibbs conditional distribution; its entries and its action on observables are given in (4). (4)✦
Each site is updated at rate one in the heat-bath generator , whose exponential kernel , for a real parameter , is defined in (5). Here is the identity matrix on . (5)
For , the random-scan kernel makes one heat-bath attempt at a uniformly chosen site.✦ Its definition is (6), with when . (6)✦
The total variation distance of a real mass function on from is one half the sum of the absolute mass differences over , as in (7). (7)✦
For a starting configuration , a real parameter , and a nonnegative integer , the continuous distance and discrete distance are the total variation distances of the rows and from the Gibbs distribution , respectively, as in (8). (8)
The Dirichlet form of a real observable on is the sum, over sites , of the mean squared deviations of from under the Gibbs distribution , as in (9). (9)✦
The Definition
Definition 1. Fix a real coupling array on sites. The continuous unmixed mass , for , is the Gibbs mass of starting configurations whose distance at remains strictly greater than , as in (10). (10)✦
The discrete unmixed mass , for , is the Gibbs mass of starting configurations whose distance after attempts remains strictly greater than , as in (11). (11)✦
For a coefficient vector , the linear spin observable is the real function on defined in (12). (12)✦
The spin covariance matrix is the real matrix of centered Gibbs spin correlations in (13). (13)✦
The covariance norm is the Euclidean operator norm of , as in (14). (14)✦
The linear relaxation time is the supremum of the variance-to-Dirichlet ratios of nonzero linear spin observables in (15), with an empty supremum assigned the value zero. (15)✦
What Is Checked
The definitions are stated over real coupling arrays on every finite spin space, with a real parameter for the exponential kernel and a nonnegative integer number of random-scan attempts.