PaperlayerView exact revision on arXiv

Source-linked AI summary

Organization of cooperation in fractal structures

Dan Peng, Ming Li

arXiv:2010.06800v2physics.soc-phnlin.AO

TL;DR

The paper asks whether fractal networks, despite their very long average distances, necessarily suppress cooperation. Using prisoner’s dilemma dynamics on fractal structures, it finds that locally compact clusters linked sparsely can facilitate cooperation, whereas fractal structures lacking such clusters can suppress it. The results suggest that both removing and inserting links can enhance cooperation relative to regular networks.

  • Problem

    Previous work commonly associated long average distance with suppressed cooperation, but fractal structures showed inconsistent effects that could not be explained by distance alone.

    Source: §I · p. 1

  • Method

    The paper studies prisoner’s dilemma evolution on hierarchical and percolation-threshold fractal networks, varying structural connectivity and local cluster organization.

    Source: §I · p. 1, §III · p. 2, §IV · p. 4

  • Results

    Fractal structures with compact clusters and sparse inter-connections can facilitate cooperation relative to regular networks, while fractal structures lacking compact clusters suppress it.

    Source: §III · pp. 2–3, §IV · p. 4, §V · p. 4

  • Takeaways & Limitations

    Cooperation can be enhanced by optimizing local connections, so a global-scale communication channel is not strictly necessary.

    Source: §V · p. 5

Abstract

from arXiv · show

It is known that the small-world structure constitutes sufficient conditions to sustain cooperation and thus enhances cooperation. On the contrary, the network with a very long average distance is usually thought of as suppressing the emergence of the cooperation. In this paper we show that the fractal structure, of which the average distance is very long, does not always play a negative role in the organization of cooperation. Compared to regular networks, the fractal structure might even facilitate the emergence of cooperation. This mainly depends on the existence of locally compact clusters. The sparse inter-connection between these clusters constructs an asymmetric barrier that the defection strategy is almost impossible to cross, but the cooperation strategy has a not too small chance. More generally, the network need not to be a standard fractal, as long as such structures exist. In turn, when this typical structure is absent, the fractal structure will also suppress the emergence of the cooperation, such as the fractal configuration obtained by diluting a random tree-like network. Our findings also clarify some contradictions in the previous studies, and suggest that both removing and inserting links from/into a regular network can enhance cooperation.

I. INTRODUCTION

Network topology shapes cooperation, but the role of long average distance remains mixed: fractal structures can differ from regular networks because of self-similarity and local organization. This paper studies whether fractal structure itself influences cooperation beyond distance alone.

I. INTRODUCTION

  • Fractal networks have power-law average distance and self-similar structure, distinguishing their dynamics from regular lattices with similarly long distances.Small-world networks instead have average distance proportional to log N and are associated with enhanced spreading and cooperation.Source: §I · p. 1
  • Prior studies report conflicting effects of long average distance, with some finding reduced cooperation and others finding enhancement through long-range links or diluted connectivity.A fractal ER giant component at average degree 1 supports no surviving cooperators, whereas percolation-threshold systems can optimize cooperation.Source: §I · p. 1
  • The paper uses the prisoner’s dilemma game to test how structural features of fractal networks organize cooperation beyond the effect of long average distance.It compares results on two different fractal structures.Source: §I · pp. 1–2
  • The hierarchical lattice is introduced as a network that can display both fractal and small-world characteristics through probabilistic preservation of long-range links during lattice growth.Its construction expands normal links into diamond-type lattices while retaining original links with probability p.Source: Figure 1 · p. 2

II. COOPERATION EVOLUTION

The cooperation dynamics use a standard prisoner’s dilemma payoff structure and stochastic strategy adoption on network neighbors. Payoffs are accumulated from neighbors, and an uncertainty parameter controls adoption noise.

II. COOPERATION EVOLUTION

  • Individuals are initially assigned cooperation or defection uniformly, receive prisoner’s dilemma payoffs from adjacent strategies, and attempt strategy transmission to randomly chosen neighbors.Cooperators gain 1 and defectors gain b when interacting with cooperators, while interactions with defectors yield 0.Source: §II · p. 2
  • Strategy adoption depends on the total payoffs of the interacting individuals, with κ = 0.1 fixing the uncertainty level in the simulations.Payoffs are recalculated after strategy changes, and one Monte Carlo step contains N update attempts.Source: §II · p. 2

III. HIERARCHICAL LATTICE

The hierarchical lattice combines fractal and small-world regimes through probabilistic long-range links, while its compact clusters and sparse interconnections shape cooperation dynamics. In the fractal limit, asymmetric strategy invasion can sustain cooperation but also greatly delay convergence.

III. HIERARCHICAL LATTICE

  • The hierarchical lattice is generated recursively from a single link, preserving each original link as a long-range connection with probability p.At p = 0 it is a large-world fractal lattice; at p = 1 it has small-world structure and high clustering.Source: §III · p. 2
  • Compared with regular lattices that reach full defection under the tested parameters, both large p and the fractal limit can support high cooperation.For b = 1.05, the fractal case can approach a nearly full C state.Source: §III · p. 2
  • The barrier stabilizes cooperative clusters because a defector crossing from a D-linked degree-2 node receives much less payoff than its cooperative neighbor inside the C cluster.The invasion probability is wb,m = [1 + e^(m−b)/κ]^-1, which becomes very small for large m.Source: §III · p. 2
  • Compact clusters linked sparsely through degree-2 nodes create an asymmetric barrier: D rarely invades C clusters, whereas C retains a substantial chance to invade D clusters.For b = 1.05 and m = 3, w1,b ≈ 0.38 while wb,m ≈ 3.4 × 10^-9.Source: §III · pp. 2–3
  • The cooperation-favoring asymmetry disappears when b is large or when strategy updates toward higher-payoff neighbors are disallowed.The restricted-update case is shown in Fig. 2(b), and high b values b = 1.5 and b = 1.75 no longer facilitate cooperation.Source: §III · p. 3, Figure 2 · p. 3
  • As p decreases, convergence becomes slower because long distances and sparse cluster interconnections impede strategy spreading.For p = 0, convergence may not occur within an acceptable time, and step-like increases correspond to complete conversion of D clusters.Source: §III · p. 3, Figure 3 · p. 3
  • Despite the delayed dynamics, p = 0 retains a non-vanishing cooperator density rather than reaching the full-D state.For sufficiently small p, the authors lack firm numerical evidence about the eventual large-system limit because of the high time complexity.Source: §III · p. 3
  • The compact-cluster mechanism is distinct from community structure: compact clusters are small local structures, whereas communities scale with system size and may contain both strategies.The paper notes that small compact clusters often contain only one strategy.Source: §III · p. 3

IV. NETWORKS NEAR THE PERCOLATION THRESHOLD

Networks near the percolation threshold combine locally compact clusters with sparse inter-connections, and this structure can enhance cooperation without requiring hubs.

IV. NETWORKS NEAR THE PERCOLATION THRESHOLD

  • Near the percolation threshold, sparse inter-connections join locally compact clusters, reproducing the structural feature associated with enhanced cooperation.Bond percolation randomly removes a fraction 1 − q of links, and square-lattice percolation has q_c = 1/2.Source: §IV · p. 4
  • Removing links from a regular lattice can enhance cooperation, with cooperator density optimized at or slightly above q_c = 1/2.This enhancement is observed for q > q_c; below q_c, the network fragments and the state is largely determined by initial strategies.Source: §IV · p. 4, Figure 5 · p. 4
  • If individuals cannot impose strategies on higher-payoff neighbors, the optimization near q_c disappears, although link removal can still produce more cooperation than q = 1.Under that update rule, cooperator density decreases monotonically with q.Source: §IV · p. 4, Figure 5 · p. 4
  • The cooperation-enhancing effect does not depend on hubs and persists across a broad range of q, rather than only at the exact fractal threshold.The contrasting fractal structures differ in whether they contain hubs, while high cooperator density occurs over a wide q range.Source: §IV · pp. 3–4
  • Fractal structure alone is insufficient: the percolating ER giant component at average degree 1 lacks locally compact clusters and cannot sustain cooperation.This limitation is consistent with cooperation peaking at a larger average degree in prior ER-network results.Source: §IV · p. 4

V. CONCLUSION

The paper attributes enhanced cooperation in fractal structures to compact cooperator-supporting clusters separated by sparse links that create asymmetric barriers to strategy spread. It concludes that long average distance need not suppress cooperation, provided the relevant local structure exists.

V. CONCLUSION

  • Locally compact clusters provide sites where cooperators can form, while sparse inter-connections make defection almost unable to cross but leave cooperation a non-negligible chance.This mechanism is stated for b values that are not too large.Source: §V · p. 4
  • The compact clusters are smaller than communities, and sparse inter-connections between them produce long convergence times for both cooperation and defection dynamics.The paper distinguishes these small clusters, often occupied by one strategy, from macro-level communities where strategies can coexist.Source: §V · p. 5
  • Both removing and inserting links into regular networks can enhance cooperation, so a global-scale communication channel is not strictly necessary.The paper instead points to optimizing local connections as a route to enhanced cooperation.Source: §V · p. 5
Loading 2010.06800v2…