Topologically Inspired Walks on Randomly Connected Landscapes With Correlated Fitness

摘要

Strictly adaptive walks on uncorrelated and correlated fitness landscapes have been a subject of intense research. However, some experimental findings tend to advance the notion of non-adaptive evolution in terms of epistasis. To address such evolutionary paths, herein we introduce the concept of topologically inspired walks on connected and correlated landscapes with complex topologies. These walks are dictated solely by the topology of connections and are not explicitly dependent on the underlying fitness values. In the biologically significant regime of sparse randomness, we observe that such topologically inspired walks might carry a population to a local optimum even faster than strictly adaptive walks. This effect becomes more pronounced with increasing correlations in fitness. We observe interesting tradeoffs between topologically inspired walks governed by the minimum and maximum value of a set of given network metrics.