A Convergence Theory for a Class of Nonlinear Programming Problems (Q5675554)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3409144
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | A Convergence Theory for a Class of Nonlinear Programming Problems |
scientific article; zbMATH DE number 3409144 |
Statements
A Convergence Theory for a Class of Nonlinear Programming Problems (English)
0 references
1973
0 references
0.797351062297821
0 references
In this paper we study a class of nonlinear programming problems of the form \[ \min \{f(x): x\in \{y\in D: g_j(y)\le 0,\ j=1,\ldots,m\}\} \] where \(f\), \(g_j: D\subset R^n\to R^1\) are continuously differentiable on the open convex set \(D\). In particular, descent methods of the type \[ x^{k+1} =x^k - \omega_k \tau_k s^k, \quad f(x^k)\ge f(x^{k+1},\ k = 0, 1,\ldots, \] with a suitable choice of direction vectors \(s^k\), steplengths \(\tau_k\), and relaxation parameters \(\omega_k\) are considered. A recent convergence theory of Elkin concerning such methods for unconstrained minimization is extended to the above nonlinear programming problem. As in Elkin's original approach, the analysis of a variety of steplength algorithms is treated entirely separately from that of several direction algorithms. This allows for their combination into many different methods for solving the constrained problem. These include some of the methods of Rosen and Zoutendijk. We also extend the results of Topkis and Veinott to non-convex sets and drop their requirement of the uniform feasibility of a subsequence of the search directions.
0 references