摘要
在时变网络图中,研究push-sum算法在量化情形下对于分布式优化问题的收敛情况,并且个体所持有的局部目标函数是强凸函数.基于算法本身的更新规则进行理论推导给出收敛分析.在一个多个体网络结构中,考虑每个节点之间只能交换量化过后的信息,采用均匀量化的方式进行探究.通过理论给出收敛性分析,并说明在量化情况下产生何种影响.经过证明得到每个节点的状态收敛到最优解附近.
We investigate the convergence rate of the push-sum Algorithm for distributed optimization over time-varying graphs.Each node maintains a strong convex function.Based on the rules of algorithm,this paper infers a convergence analysis.This paper studies the problem of optimizing the sum of the individual local strong convex functions which are know to each agent accordingly.Each agent only exchanges quantized information with its neighbors.The state of each node converges to the vicinity of the optimal solution.
作者
黄继英
HUANG Jiying(School of Mathematicl Science,Chongqing Normal University,Chongqing 401331,China)
出处
《湖北民族学院学报（自然科学版）》
CAS
2018年第3期326-334,共9页
Journal of Hubei Minzu University(Natural Science Edition)
关键词
量化
强凸优化
分布式
push-sum算法
quantization
strong conwex optimization
distributed optimization
push-sum algorithm
