Communication in complex networks

One of the properties interest in analysis networks is global communicability, i.e., how easy or difficult it is, generally, to reach nodes from other by following edges. Different communicability measures provide quantitative assessments this property, emphasizing different aspects problem. This paper investigates sensitivity local changes. In particular, for directed, weighted networks, we study change when weight a single edge changed; or, unweighted case, an added removed. The include total network based on matrix exponential adjacency matrix, and Perron defined terms root associated left right eigenvectors. Finding what changes lead largest has many potential applications, including assessing resilience system failure attack, guidance incremental improvements, studying errors connection data.