Agent Based Models

Resilience of Single-Layer and Multiplex Networks Following Sudden Changes to Tie Costs

Author: Paul Smaldino

An agent-based model of incentive-based tie formation in a two-layer multiplex network. The incentives include costs and benefits for direct ties, closed triangles, and spillover edges (edges that exist in both layers of the multiplex). The network is subject to shocks: sudden changes to tie costs or other incentives after the process of network formation has commenced.


Empirically Grounded Simulation of Multiplex Co-Evolution

Author: Camber Warren

This project presents an ABM simulation framework for analyzing the dynamic co-evolution of multiplex networks, characterized by arbitrary numbers of edge types and node attributes.  The framework is based on an agent-level utility function, which consists of weighted preferences for various network characteristics, including transitivity, homophily, preferential attachment, and spillover effects between network edge types.  The library includes functions for empirical estimation of the parameters of the utility function using constrained sequential least-squares programming.  It also includes functions for estimating and visualizing the response of multiplex networks to random shocks, given specified utility coefficients, where shocks consist of exogenous events of node addition, node deletion, and the creation of edge barriers that prohibit ties between particular node pairs.