= Percolation triangle condition
{title2=$\sum_{x,y}\tau(0,x)\tau(x,y)\tau(y,0)<\infty$}
For <independent> <bond percolation> on $\mathbb Z^d$, let $\tau(x,y)=\mathbb P_{p_c}(x\leftrightarrow y)$ be the critical <percolation two-point connection probability>. The <percolation triangle condition> is
$$
\nabla(p_c)=\sum_{x,y}\tau(0,x)\tau(x,y)\tau(y,0)<\infty.
$$
<translation invariance> makes this a threefold discrete <convolution> evaluated at zero. The factors describe connections between three positions; their product is a diagrammatic weight, rather than the <probability> of three <independent> events in a single configuration.
The condition controls the intersection corrections that separate <percolation clusters> from a simple branching approximation. In the homogeneous setting, it implies the <mean-field percolation exponents> value $\gamma=1$ for the divergence of <percolation susceptibility>. It must be established for the particular model; <lace expansion> provides one method in suitable high-dimensional models. The condition is therefore a useful bridge between spatial connectivity estimates and critical behaviour, rather than an assumption automatically satisfied at criticality.<Tree graph inequalities and critical behavior in percolation models by Michael Aizenman and Charles M. Newman>[Triangle convergence criterion and susceptibility exponent; original paper, Journal of Statistical Physics 36 (1984), pages 107–143]{ref}
Back to article page