Janus Percolation in Anisotropic Limited-Degree Networks

Abstract

Many real-world infrastructures, from sensor and road networks to power grids, are spatially embedded and anisotropic, with constraints on the maximum number of links each node can establish. Such systems can be represented as anisotropic limited-degree networks, in which each node forms at most q outgoing links preferentially oriented along a fixed direction. By increasing the node density sigma at fixed q, we uncover a reentrant percolation transition: a giant strongly connected component emerges, but unexpectedly disintegrates again at high densities. This counterintuitive behavior implies that adding nodes, normally expected to enhance robustness, can instead reduce mutual accessibility and weaken global connectivity. The critical behavior displays two coexisting "faces": random-percolation scaling along the preferred direction and directed-percolation scaling transversely, therefore we name this phenomenon Janus percolation, in analogy with the dual-faced Roman god. These findings demonstrate that anisotropy and degree limitation can jointly induce a novel reentrant connectivity with mixed universality that bridges the universality classes of random and directed percolation, providing fresh insight into how structural constraints shape connectivity and resilience in spatial networks.

Figures (4)

• Figure 1: From isotropic to anisotropic connectivity with increasing density. Schematic illustration of the model for a maximum out-degree $q=2$, at density $\sigma=0.3$ (a) and $\sigma=0.6$ (b). Each node $i$ connects to up to $q$ neighboring nodes located within a distance $2r$, preferentially oriented downward along a fixed vertical direction (e.g node 0 connects to nodes 3 and 4, in panel a). We show a low density configuration in panel (a), and higher density of nodes in panel (b). Increasing density, more neighbors become available, allowing nodes to connect further along the preferred vertical direction (e.g. node 2 switching connection from node 0 in (a) to 11 in (b)), thus enhancing anisotropy. Connectivity is analyzed in terms of strongly and weakly connected components (SCCs and WCCs): in SCCs (orange) every node can reach every other by following directed links, whereas in WCCs (orange and blue) the same definitions holds when links are treated as undirected. Note that in panel (b), because of the strong anisotropy, SCC disappears.
• Figure 2: Reentrant percolation phase diagram of strongly connected components. Phase diagram in the $(\sigma, q)$ plane, where $\sigma$ is the node density and $q$ the maximum out-degree. The solid line marks the transition between the non-percolating and percolating phases, where the black and green colors are used to denote RP and JP universality classes, respectively. As $\sigma$ increases at fixed $q \gtrsim 4.7$, a giant strongly connected component (SCC) first emerges (percolating phase, orange) but this fragments again at higher densities (non-percolating phase, blue), revealing a reentrant transition. Insets show representative snapshots for a system of size $L=1600$, where colors denote distinct SCCs; only components with size $M>500$ are displayed. The vertical dashed line corresponds to the critical threshold ($\sigma_\mathrm{B}$) of the Boolean model.
• Figure 3: Finite-size scaling of correlation lengths revealing a JP transition. Filled and open symbols represent correlation lengths measured along the direction transverse (top panels) and parallel (bottom panels) to the preferred direction, respectively. Panels (a) and (b) show the raw correlation lengths for strongly connected components (SCCs) at $q=7$, the vertical dashed lines are the thresholds for the first and second transition. The corresponding data collapses are presented in the right column: (c,d) are respectively the collapses for the first and second transition of panel (a), similarly (e,f) are those for data in panel (b). Collapses are obtained using the finite-size scaling Ansatz of Eq. \ref{['scaling_ansatz']}, with $\epsilon = (\sigma - \sigma_{c_i})/\sigma_{c_i}$ with $i=1$ for panels (c,e) and $i=2$ for panels (d,f). The best data collapses yield $\sigma_{c_1} = 0.3615$ and exponents $\nu_{p} = \nu_{t}= 4/3$ for the first transition (panels (c,e)), and $\sigma_{c_2} = 1.058$ with $\nu_{p} = 1.734$ and $\nu_{t}=4/3$ for the second transition (panels (d,f)).
• Figure 4: Anisotropy degree of strongly connected components as a function of node density. The anisotropy degree $\eta = G_{t}^2 / G_{p}^2$ quantifies the elongation of strongly connected components (SCCs), where $G_{t}$ and $G_{p}$ are the projected gyration radii along the transverse and parallel directions, defined in Eq. \ref{['gyration']}.The vertical dashed lines $\sigma_\mathrm{c_1}$ and $\sigma_\mathrm{c_2}$ are the thresholds for the first and second transitions. The top panel shows results for $q \ge 6$, where the first percolation transition is nearly isotropic, with $\eta \simeq 1$ (RP), while the second transition exhibits strong anisotropy with elongated SCCs (JP). The bottom panel corresponds to $4.7 \lesssim q < 6$, for which both transitions are JP. In all of these cases, $\eta$ reaches a pronounced maximum near the percolation threshold, signaling the formation of elongated clusters transverse to the preferred direction.