© 1970 by British Computer Society
An algorithm for solving nonlinear programming problems subject to nonlinear inequality constraints
Advanced Structures Group, Allison Division, General Motors Corporation, Indianapolis, Indiana, USA
An algorithm of the penalty function class which solves linear or non-linear optimisation problems subject to equality and/or inequality constraints is described. The penalty term consists of exponential summands like exp [T·g(x)] where T < 0 and g(x)
0 is a constraint. Computational experience is discussed. Convergence is proved and is typically first-order. The algorithm has found considerable application in its ability to distinguish readily between feasible and non-feasible (i.e. no-domain) problems.
Received November 1968.
* Advanced Structures Group, Allison Division, General Motors Corporation, Indianapolis, Indiana, U.S.A.