Network-Configurations of Dynamic Friction Patterns

The complex configurations of dynamic friction patterns-regarding real time contact areas- are transformed into appropriate networks. With this transformation of a system to network space, many properties can be inferred about the structure and dynamics of the system. Here, we analyze the dynamics of static friction, i.e. nucleation processes, with respect to “friction networks”. We show that networks can successfully capture the crack-like shear ruptures and possible corresponding acoustic features. We found that the fraction of triangles remarkably scales with the detachment fronts. There is a universal power law between nodes’ degree and motifs frequency (for triangles, it reads T(k)\proptok{\beta} ({\beta} \approx2\pm0.4)). We confirmed the obtained universality in aperture-based friction networks. Based on the achieved results, we extracted a possible friction law in terms of network parameters and compared it with the rate and state friction laws. In particular, the evolutions of loops are scaled with power law, indicating the aggregation of cycles around hub nodes. Also, the transition to slow rupture is scaled with the fast variation of local heterogeneity. Furthermore, the motif distributions and modularity space of networks -in terms of withinmodule degree and participation coefficient-show non-uniform general trends, indicating a universal aspect of energy flow in shear ruptures.

💡 Research Summary

The paper introduces a novel framework—“friction networks”—that maps real‑time contact‑area measurements of sliding interfaces onto complex networks in order to quantify the dynamics of static‑to‑dynamic friction transitions, especially the nucleation and propagation of rupture fronts. The authors use both two‑dimensional optical imaging of transparent blocks and one‑dimensional line‑scan data. Each spatial profile (or patch) is treated as a node; edges are created between nodes whose correlation exceeds a threshold selected from a region of minimal edge‑density change in the betweenness‑centrality versus correlation space. Standard network metrics (degree k, clustering coefficient C, modularity M, assortativity r, within‑module degree Z, participation coefficient P) are then computed for each temporal snapshot.

A central empirical finding is that the number of triangles (3‑node loops) attached to a node scales with its degree as T(k)∝k^β with β≈2±0.4, a relationship that holds across all experiments, including aperture‑based friction data from rock samples. This scaling indicates that as a rupture front advances, loops concentrate around hub nodes, producing a highly clustered local structure superimposed on a scale‑free global topology.

The authors further hypothesize that variations in the clustering coefficient are proportional to variations in shear stress (∂σ/∂t∝∂C/∂t). By integrating this relation they obtain a logarithmic dependence of shear stress on both degree and triangle count, and they derive an explicit link between rupture‑front velocity v_front and the spatial‑temporal gradient of the triangle fraction: v_front∝(∂C/∂x)⁻¹. This provides a direct, measurable bridge between a purely topological quantity (C) and a physical observable (front speed).

Loop statistics follow a power‑law distribution P(T)∝T^−γ with γ≈2.0, confirming that the network’s loops are not uniformly distributed but instead aggregate around hubs. The assortativity coefficient is consistently positive, indicating a tendency for nodes of similar degree to connect. In the one‑dimensional data, each passage of a rupture front produces a sharp spike in r, mirroring the sudden increase in connectivity observed in acoustic emission records.

Modularity analysis reveals that the evolution of the network in the (Z,P) space follows distinct trajectories for different rupture regimes. Sub‑Rayleigh fronts generate rapid, high‑Z, low‑P excursions, whereas slow‑rupture phases are characterized by a monotonic rise in modularity with comparatively modest changes in assortativity. This suggests that energy flow becomes increasingly confined within specific modules during slow slip, while fast fronts momentarily redistribute connections across modules.

Crucially, the authors embed the network descriptors into a conventional rate‑and‑state friction law. The state variable θ, traditionally governed by Ruina’s evolution equation (∂θ/∂t = −θ·V/D), is expressed as a linear combination of local loop count and global degree: θ≈a·T + b·k. Differentiating yields ∂θ/∂t≈a·∂T/∂t + b·∂k/∂t, which, together with the empirical T(k) scaling and the C–σ relation, reproduces the standard state‑evolution dynamics while explicitly accounting for the topological re‑organization of the interface. When the network approaches a “gel‑like” state—where a single hub connects to almost all other nodes—the model predicts a rapid drop in effective friction, consistent with the observed transition to slow rupture.

Overall, the study demonstrates that transforming frictional contact patterns into network representations uncovers a suite of robust, scale‑invariant signatures: (1) triangle‑degree power law, (2) clustering‑stress proportionality, (3) loop power‑law distribution, (4) assortativity spikes at front passages, and (5) modularity trajectories that differentiate rupture modes. These signatures not only validate the network approach across different materials and measurement techniques but also provide a concrete pathway to augment classical friction laws with measurable topological parameters, thereby offering a richer, multiscale description of fault slip and laboratory shear experiments.