Distributed primal-dual method for multi-agent sharing problem with conic constraints
Distributed primal-dual method for multi-agent sharing problem with conic constraints
复制标题
DOI：
10.1109/acssc.2016.7869152
复制
发表时间：
2016-11
期刊：
影响因子：
--
通讯作者：
N. Aybat;E. Y. Hamedani
中科院分区：
文献类型：
--
作者：
N. Aybat;E. Y. Hamedani
我们考虑了无向代理网络上的协作多代理资源共享问题，在该网络中，只有被边连接的代理才能直接通信。目标是最小化代理特定的复合凸函数之和，该函数受耦合代理决策的圆锥约束的约束。提出了一种求解鞍点问题的分布式原始-对偶算法，该算法需要计算耦合约束的一致对偶代价。我们在次最优、不可行和违反共识的情况下给出了代理双重价格评估的收敛速度；考察了底层网络拓扑结构对所提出的分散算法的收敛速度的影响；并在基追踪问题上与Prox-JADMM算法进行了比较。
We consider cooperative multi-agent resource sharing problems over an undirected network of agents, where only those agents connected by an edge can directly communicate. The objective is to minimize the sum of agent-specific composite convex functions subject to a conic constraint that couples agents' decisions. A distributed primal-dual algorithm is proposed to solve the saddle point formulation, which requires to compute a consensus dual price for the coupling constraint. We provide convergence rates in sub-optimality, infeasibility and consensus violation for agents' dual price assessments; examine the effect of underlying network topology on the convergence rates of the proposed decentralized algorithm; and compare our method with Prox-JADMM algorithm on the basis pursuit problem.