Theory of Boundary Effects in Invasion Percolation

TL;DR

This paper investigates how boundaries influence invasion percolation, revealing new surface critical exponents, fractal dimension differences near boundaries, and a logarithmic crossover, with theoretical and numerical analyses showing boundary effects mainly alter geometrical properties.

Contribution

It introduces a theoretical framework combining Fixed Scale Transformation and Run Time Statistics to analyze boundary effects in invasion percolation, providing new critical exponents and insights.

Findings

Boundary introduces new surface critical exponents.

Fractal dimension near boundary differs from bulk.

Theoretical predictions agree well with numerical data.

Abstract

We study the boundary effects in invasion percolation with and without trapping. We find that the presence of boundaries introduces a new set of surface critical exponents, as in the case of standard percolation. Numerical simulations show a fractal dimension, for the region of the percolating cluster near the boundary, remarkably different from the bulk one. We find a logarithmic cross-over from surface to bulk fractal properties, as one would expect from the finite-size theory of critical systems. The distribution of the quenched variables on the growing interface near the boundary self-organises into an asymptotic shape characterized by a discontinuity at a value $x_c=0.5$, which coincides with the bulk critical threshold. The exponent $\tau^{sur}$ of the boundary avalanche distribution for IP without trapping is $\tau^{sur}=1.56\pm0.05$; this value is very near to the bulk one. Then we conclude that only the geometrical properties (fractal dimension) of the model are affected by the presence of a boundary, while other statistical and dynamical properties are unchanged. Furthermore, we are able to present a theoretical computation of the relevant critical exponents near the boundary. This analysis combines two recently introduced theoretical tools, the Fixed Scale Transformation (FST) and the Run Time Statistics (RTS), which are particularly suited for the study of irreversible self-organised growth models with quenched disorder. Our theoretical results are in rather good agreement with numerical data.