On the finite e-convergence of exterior penalty methods

Abstract EN

K.

Truemper [1975] proved that it is possible to select a target value fx* > 0 for the penalty parameter of the exterior function Pr (x, jx) = f{x) + jx X™ 1 max(0, -g, (*))' where r = 2 such that when jx is increased to this value or becomes larger, then x which minimizes Pr(x,\i*) is within a specified tolerance from the optimal solution of the problem mmf(x) subject to g[(x) > 0, Vi = \,...,m.

In this paper, we extend his results for r > 1.