Optimizing synchronizability of networks

In this paper, we investigate the factors that affect the synchronization of coupled oscillators on networks. By using the edge-intercrossing method, we keep the degree distribution unchanged to see other statistical properties' effects on network's synchronizability. By optimizing the eigenratio $R$ of the coupling matrix with \textit{Memory Tabu Search} (MTS), we observe that a network with lower degree of clustering, without modular structure and displaying disassortative connecting pattern may be easy to synchronize. Moreover, the optimal network contains fewer small-size loops. The optimization process on scale-free network strongly suggests that the heterogeneity plays the main role in determining the synchronizability.