CONVERGENCE THEORIES OF DISTRIBUTED ITERATIVE PROCESSES : A SURVEYt by

We consider a model of distributed iterative algorithms whereby several processors participate in the computation while collecting, possibly stochastic information from the environment or other processors via communication links. Several applications in distributed optimization, parameter estimation, and communication networks are described. Issues of asymptotic convergence and agreement are explored under very weak assumptions on the ordering of computations and the timing of information reception. Progress towards constructing a broadly applicable theory is surveyed. tThe research of D.P. Bertsekas was supported by NSF-ECS-8217668 and under DARPA Grant ONR-N00014-75-C-1183. The research of J.N. Tsitsiklis and M. Athans was supported by ONR-N00014-77-C-0532(NR041-519) and ONR N00014-84-K-0519. *Dept. of Electrical Engineering and Computer Science, Laboratory for Information and Decision Systems, M.I.T., Cambridge, MA. 02139.