Percolation in an ultrametric space (Q388850)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Percolation in an ultrametric space
scientific article

    Statements

    Percolation in an ultrametric space (English)
    0 references
    0 references
    0 references
    17 January 2014
    0 references
    This paper studies percolation on a metric space which is ultra-metric. That is, if \(d(\cdot, \cdot)\) denotes the metric function, then for any three points \(x,y\) and \(z\) we have \(d(x,y) \leq \max \{ d(x,z), d(y,z) \}\). More specifically, the space consists of all infinite sequences of the form \((x_1, x_2, \ldots)\) where \(x_i \in \mathbb{Z}_N\) and \(\sum_i x_i < \infty\). Here, \(N\) is a fixed parameter. This space is denoted by \(\Omega_N\). Subsequently, a random graph is formed where two points \(x,y\) are joined by an edge independently of any other pair with probability that depends on their distance: if \(d(x,y) = k\), then this probability is equal to \(\min \{ c_k / N^{k(1+\delta)} ,1\}\), where \(\delta > - 1\) and \(c_k>0\). The main question that is considered here has to do with whether or not percolation occurs at an arbitrarily chosen point. Consider the point \(\mathbf{0}\), which is the all-zeros element of \(\Omega_N\). The paper gives sufficient conditions for the parameters \(c_k\) and \(\delta\) so that the probability that \(\mathbf{0}\) belongs to an infinite cluster is positive. If the latter is the case, then we say that percolation occurs. It seems that \(\delta = 1\) is the critical value for this. If \(\delta > 1\) but \(\sup_k c_k < \infty\), then percolation does not occur. If \(\delta < 1\) and \(c = \inf_k c_k\) is sufficiently large, then percolation occurs and the infinite component is unique. For the \(\delta = 1\) case, the authors consider \(c_k\)s of a specific form: \(c_k = C_0 + C_1 \log k + C_2 k^{\alpha}\), with \(\alpha \geq 0\) and also \(C_0,C_1,C_2 \geq 0\). The situation here is mixed and depends on the combined values of these four parameters. If \(C_2=0\) and \(C_1 < N\), then, no matter what the values of \(\alpha\) and \(C_0\) are, percolation does not occur. The remaining results assume that \(\alpha > 2\). Here, percolation may or may not occur depending on certain combinations of the values of \(C_0,C_1, C_2\).
    0 references
    percolation
    0 references
    ultrametric spaces
    0 references
    infinite cluster
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references