Search for: [Abstract = "The developements of the theory of augmented Lagrange functions \(or, equivalently, shifted penalty functions\) resulted in powerful saddle\-point theorems and, recently, in new single\-loop iterative algorithms for saddle\-point seeking and for solving optimization problems with constraints. A particularly strong new algorithm is based on two variable metric approximations and a constraint shift \(or violation\) prediction. For large scale optirnization, this algorithm leads to a primal\-dual coordination method. The method converges in a finite member of steps for quadratic problems with linear constraints and interactions, and generally converges rapidly for more complicated problems. The method has also other advantages of shifted penalty or augmented Lagrange funktions when applied to large scale optimization. lt is also hoped that the method can be used for on\-line coordination in hierarchical control systems."]

Number of results: 1