Title :
Distributed asynchronous gradient algorithms for convex network flow problems
Author_Institution :
LAAS du CNRS, Toulouse, France
Abstract :
The author considers the solution of the single commodity strictly convex network flow problem. The dual of this problem is unconstrained and differentiable. It is shown that the structure of the dual problem allows the successful application of a distributed asynchronous gradient method whereby gradient iterations are carried out by several processors in arbitrary order and with arbitrarily large interprocessor communication delays
Keywords :
conjugate gradient methods; convex programming; economic cybernetics; convex network flow problems; distributed asynchronous gradient method; dual problem; gradient iterations; interprocessor communication delays; single commodity; Computer networks; Convergence; Cost function; Delay; Distributed algorithms; Distributed computing; Gradient methods; Iterative methods; Relaxation methods; Telecommunication traffic;
Conference_Titel :
Decision and Control, 1992., Proceedings of the 31st IEEE Conference on
Conference_Location :
Tucson, AZ
Print_ISBN :
0-7803-0872-7
DOI :
10.1109/CDC.1992.371154
