A complex system consists of many interacting units, whose collective behavior cannot be explained from the behavior of the individual units alone. Complex dynamical networks are complex systems that can be represented with graphs dynamically evolving in time. Computational grids provide new natural examples of large-scale complex networks emerging from collective behavior. Moreover, computational grids feature multiple levels of interactions. An interesting question is thus whether these networks will exhibit properties similar to those of social networks, or original ones, which would be the specific signature of e-science. An operational question is the creation of generative models appropriate for forecasting future graph structure. The PhD will characterize the spatiotemporal structure of the graphs created 1) by co-access to files, and 2) by the job traffic.