Gradient Descent for Resource Allocation with Packet Loss*

This paper studies the effect of packet loss during application weighted gradient descent to solve a resource allocation problem with piecewise quadratic cost functions in multi-agent system. We define two performance metrics that measure, respectively, deviation from constraint and error on expected function. derive upper bounds both metrics: are proportional difference between initial function evaluated at minimizer. Then, we extend analysis violation open systems where agents replaced: based preliminary result simulations show combination replacements losses makes diverge time.