Simple model for directed networks Luis G. Morelli* Abdus Salam International Centre for Theoretical Physics, P.O. Box 586, 34100Trieste, Italy Received 24 January 2003; published 17 June 2003 We study a model for directed networks based on the Watts-Stogatz model for small-world phenomena. We focus on some topological aspects of directed networks inspired in food web theory, namely, the fraction of basal and top nodes in the network and node level distributions. We argue that in directed networks basal nodes play an important role, collecting information or resources from the environment. We give analytical expres- sions for the fraction of basal and top nodes for the model, and study the node level distributions with numerical simulations. DOI: 10.1103/PhysRevE.67.066107 PACS numbers: 02.50.-r, 89.75.Hc, 05.10.-a I. INTRODUCTION The subject of complex networks has become relevant in several branches of science. Sometimes, complex systems can be described as a large set of interacting units. Regard- less of the nature of these units and the way they interact with each other, it is possible to define a network describing the interactions present in the system, thinking of the units as the nodes of the network and placing a link between two units whenever an interaction is present. The interest in com- plex networks has been triggered both by novel theoretical approaches 1–3and by the availability of large datasets for networks of different origins. In the recent literature, a sys- tematic study of the topological properties of neural 1,4, ecological 5–7, metabolic 8, transport 9, technological 1,10,11and social 1,9,12,22,23networks has been carried out, revealing a rich subject with a wide scope of applica- tions. Most of the models proposed to account for the observed features in real networks consist of undirected networks 13, i.e., networks in which the links connecting two nodes have no definite direction. However, in many cases of interest the interactions are not symmetric. A natural way to describe these networks is to consider directed links. One example of directed networks comes from ecology, where ecosystems can be represented by food webs 14. In predator-prey food webs, each species is represented as a node of the network, and a link is placed between two species whenever one of them feeds on the other. Food webs describe in this way the who-eats-who interactions in the ecosystem. The natural way to describe the flow of resources from prey to predators is to consider a directed network. 1 Neural networks provide an- other example of directed networks. In fact, neurons connect to other cells with their axon, and receive connections from other axon cells through their dendrites. This asymmetry of the cell can be accounted for with a directed link. In ener- getic networks as the power grid, the energy flows in some definite direction, from providers to consumers. Also in economy, directed networks play an important role. The net- work of goods necessary to produce other goods is an ex- ample. Hierarchical social networks, in which the relation- ships between people are not symmetric, give yet another example of directed networks. The relevance of directed networks has already been rec- ognized and emphasized 13,17. A model for the World Wide Web with directed links has been proposed 16and a spreading process occurring on a directed network has been studied 18. The problem of percolation in a directed scale- free network 19and the dynamics of a spin model in a directed small world network 20have also been consid- ered. In this paper, we propose a model for directed networks based on the Watts-Stogatz model for small-world phenom- ena. It is a simple extension of the model, modified to ac- count for the direction of the links. We focus our study on some static topological properties inspired in food web theory 15. In the following section, we introduce some definitions concerning the topology of directed networks. In the following, we present the model and give analytical ex- pressions for the fraction of basal and top nodes. Then we show numerical results for the fraction of basal nodes and the node level distribution. We finish the paper with a discussion of the results and some remarks. II. SOME ASPECTS OF FOOD WEB TOPOLOGY We therefore begin with a brief discussion of some as- pects of network topology inspired in food web theory. A very coarse grained classification of the species in a food web is based on the fraction of basal species, the fraction of top species, and the fraction of intermediate species in the web. Basal species feed only on the environment, and have no prey. Top species have no predators. Intermediate species have both preys and predators. Here we define basal, top, and intermediate nodes in an analogous way. Basal nodes have only outgoing links, intermediate nodes have both outgoing and incoming links, and top nodes have only incoming links. The fractions of basal, intermediate, and top nodes in the network are noted as B, I, and T. Note that by definition, B +T +I =1. In food web models, basal species feed only on the envi- ronment. The ecosystem is an open system, and resources *Electronic address: morelli@ictp.trieste.it 1 In some contexts, though, food webs can be represented by un- directed networks because a fluctuation in the population of a spe- cies has consequences both on the populations of its preys and its predators. PHYSICAL REVIEW E 67, 066107 2003 1063-651X/2003/676/0661077/$20.00 ©2003 The American Physical Society 67 066107-1