Core Concepts
Optimizing Markov chain connectivity in weighted graphs under uncertainty.
Abstract
Weighted graphs model complex systems like social networks, power grids, and transportation networks. Mean First Passage Times (MFPTs) characterize network connectivity. Optimizing MFPTs using Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm improves scalability. Extending metrics for network connectivity based on MFPTs allows for more general objective functions. The impact of reversibility on optimal solutions is explored. A gradient-based optimization method is proposed for non-convex network design problems.
Stats
Mean First Passage Times (MFPTs)
Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm
Quotes
"Our relatively simple yet powerful extension of SPSA underscores the versatility of first-order methods in studying network connectivity."