Driven aging dynamics on sparse networks
Benedikt Grüger (University Göttingen)

We investigate the interaction of two paradigmatic ways of being out of equilibrium, aging and driving, in simple models of glassy dynamics. We specifically consider the Bouchaud model, where a system jumps between the numerous minima of a rough energy landscape in configuration space. As the temperature decreases, the system undergoes a dynamical phase transition, at which the relaxation time diverges. With an additional field, we then drive the system by biasing it's dynamics towards higher/lower jumping activity. We investigate the spectrum of the (biased) master operator in that framework, using a population dynamics algorithm based on cavity theory that allows us to deduce statements about the thermodynamic limit. Combining this with extensive diagonalization we identify novel regimes in the bias-temperature phase diagram that are distinguished by the occurrence of different kinds of eigenvector localization and are linked to the existence of a spectral gap. We also present methodological advances in the form of novel strategies for operating the population dynamics algorithm.