We consider a variant of the gossip algorithm wherein a controller can influence the values at a subset of nodes as well as the averaging weights. This leads to three optimization problems: (i) Optimal choice of nodes: a combinatorial optimization problem for which we propose a nonlinear programming relaxation as well as a greedy heuristic; (ii) Parametric optimization of weights: a non-convex optimization for which we propose an effective heuristic for a special case; (iii) Dynamic adjustment of weights: an optimal control problem. For the dynamic case, we note some empirically observed interesting critical phenomena for the uncontrolled case.
No takes yet. Share an insight, caveat, or question.
Borkar et al. (2014) studied this question.
Synapse has enriched 2 closely related papers on similar clinical questions. Consider them for comparative context: