Thimo Rohlf

Topological evolution of dynamical networks: global criticality from local dynamics.

We evolve network topology of an asymmetrically connected threshold network by a simple local rewiring rule: quiet nodes grow links, active nodes lose links. This leads to convergence of the average connectivity of the network towards the critical value K(c) = 2 in the limit of large system size N. How this principle could generate self-organization in natural complex systems is discussed for two examples: neural networks and regulatory networks in the genome.

Criticality in Random Threshold Networks: Annealed Approximation and Beyond

Random Threshold Networks with sparse, asymmetric connections show complex dynamical behavior similar to Random Boolean Networks, with a transition from ordered to chaotic dynamics at a critical average connectivity Kc. In this type of model—contrary to Boolean Networks—propagation of local perturbations (damage) depends on the in-degree of the sites. Kc is determined analytically, using an annealed approximation, and the results are confirmed by numerical simulations. It is shown that the statistical distributions of damage spreading near the percolation transition obey power-laws, and dynamical correlations between active network clusters become maximal. We investigate the effect of local damage suppression at highly connected nodes for networks with scale-free in-degree distributions. Possible relations of our findings to properties of real-world networks, like robustness and non-trivial degree-distributions, are discussed.

Damage spreading and criticality in finite random dynamical networks.

We systematically study and compare damage spreading at the sparse percolation (SP) limit for random Boolean and threshold networks with perturbations that are independent of the network size N. This limit is relevant to information and damage propagation in many technological and natural networks. Using finite-size scaling, we identify a new characteristic connectivity Ks, at which the average number of damaged nodes d[over ], after a large number of dynamical updates, is independent of N. Based on marginal damage spreading, we determine the critical connectivity Kc(sparse)(N) for finite N at the SP limit and show that it systematically deviates from Kc, established by the annealed approximation, even for large system sizes. Our findings can potentially explain the results recently obtained for gene regulatory networks and have important implications for the evolution of dynamical networks that solve specific tasks.

Emergent Criticality Through Adaptive Information Processing in Boolean Networks

We study information processing in populations of boolean networks with evolving connectivity and systematically explore the interplay between the learning capability, robustness, the network topology, and the task complexity. We solve a long-standing open question and find computationally that, for large system sizes N, adaptive information processing drives the networks to a critical connectivity K(c)=2. For finite size networks, the connectivity approaches the critical value with a power law of the system size N. We show that network learning and generalization are optimized near criticality, given that the task complexity and the amount of information provided surpass threshold values. Both random and evolved networks exhibit maximal topological diversity near K(c). We hypothesize that this diversity supports efficient exploration and robustness of solutions. Also reflected in our observation is that the variance of the fitness values is maximal in critical network populations. Finally, we discuss implications of our results for determining the optimal topology of adaptive dynamical networks that solve computational tasks.

Long-range memory elementary 1D cellular automata: Dynamics and nonextensivity

We numerically study the dynamics of elementary 1D cellular automata (CA), where the binary state σi(t)∈{0,1} of a cell i does not only depend on the states in its local neighborhood at time t-1, but also on the memory of its own past states σi(t-2),σi(t-3),…,σi(t-τ),… . We assume that the weight of this memory decays proportionally to τ-α, with α⩾0 (the limit α→∞ corresponds to the usual CA). Since the memory function is summable for α>1 and nonsummable for 0⩽α⩽1, we expect pronounced changes of the dynamical behavior near α=1. This is precisely what our simulations exhibit, particularly for the time evolution of the Hamming distance H of initially close trajectories. We typically expect the asymptotic behavior H(t)∝t1/(1-q), where q is the entropic index associated with nonextensive statistical mechanics. In all cases, the function q(α) exhibits a sensible change at α≃1. We focus on the class II rules 61, 99 and 111. For rule 61, q=0 for 0⩽α⩽αc≃1.3, and q αc, whereas the opposite behavior is found for rule 111. For rule 99, the effect of the long-range memory on the spread of damage is quite dramatic. These facts point at a rich dynamics intimately linked to the interplay of local lookup rules and the range of the memory. Finite size scaling studies varying system size N indicate that the range of the power-law regime for H(t) typically diverges ∝Nz with 0⩽z⩽1.

Self-organized pattern formation and noise-induced control based on particle computations

We propose a new non-equilibrium model for spatial pattern formation based on local information transfer. Unlike most standard models of pattern formation it is not based on the Turing instability or initially laid down morphogen gradients. Information is transmitted through the system via particle-like excitations whose collective dynamics results in pattern formation and control. Here, a simple problem of domain formation is addressed by means of this model in an implementation as stochastic cellular automata, and then generalized to a system of coupled dynamical networks. One observes stable pattern formation, even in the presence of noise and cell flow. Noise contributes through the production of quasi-particles to de novo pattern formation as well as to robust control of the domain boundary position. Pattern proportions are scale independent as regards system size. The dynamics of pattern formation is stable over large parameter ranges, with a discontinuity at vanishing noise and a second-order phase transition at increased cell flow.