Abstract:
Designing effective recovery strategies for damaged networks is important across built, human, and natural systems. Postperturbation network recovery has been motivated by two distinct philosophies, specifically, the use of centrality measures in complex networks versus network optimization measures. The hypothesis that hybrid approaches may offer complementary value and improve our understanding of recovery processes while informing real-world restoration strategies has not been systematically examined. This research shows that the two distinct network philosophies can be blended to form a hybrid recovery strategy that is more effective than either. Network centrality�based metrics tend to be intuitive and computationally efficient but remain static irrespective of the desired functionality or damage pattern. Optimization-based approaches, while usually less intuitive and more computationally expensive, can be dynamically adjusted. The proposed approach, based on edge recovery algorithms with edge importance informed by network flow and node attributes, outperforms recovery informed exclusively either by network centrality or network optimization. We find that optimization methods outperform centrality-based approaches for networks that are large enough for the power law to be manifested, but for treelike networks typically found at smaller scale, the two approaches are competitive and scenario specific.