Infection processes on networks with structural uncertainty

摘要: Over the last ten years, interest in network phenomena and potential for a global pandemic have produced tremendous volume of research exploring consequences human interaction patterns disease propagation. The often focuses on single question: will an emerging infection become epidemic? This thesis clarifies relationships among different epidemic threshold criteria deterministic models, discusses role meaning basic reproductive ratio, R0. We quantify incorporation population structure into this general framework, identify conditions under which topology characteristics can be decoupled computation functions, generalizes many existing results literature. decoupling allows us to focus impact via spectral radius adjacency matrix network. It is rare, however, that one has complete information about every disease-transmitting interaction; uncertainty ignored models. Neglecting lead underestimate R0, unacceptable outcome public health planning. Is it possible make guarantees approximations regarding spread when only partial routes transmission known? present methods making predictions over uncertain networks, including approximation techniques bounding obtained graph theory, illustrate these several data sets. also approach problem by using simulation analytical work characterize radii arise from members exponential random family, commonly used model empirical networks quantitative sociology. Finally, we explore issues spatiotemporal propagation through network, focusing behavior contact process influence model. Thesis Supervisor: George Verghese Title: Professor Electrical Engineering Computer Science