A recursive quadratic programming algorithm that uses differentiable exact penalty functions

From MaRDI portal





Considered is the equality constrained nonlinear minimization problem: minimize \(f(x)\) subject to \(c(x)=0\), where \(f: \mathbb R^ n\to \mathbb R^ 1\) and \(c: \mathbb R^ n\to R^ m\) are twice continuously differentiable functions. The recursive quadratic programming method presented here for solving the above problem uses approximations to Fletcher's differentiable exact penalty function [\textit{R. Fletcher}, ibid. 5, 129--150 (1973; Zbl 0278.90063)] as line search functions. The algorithm has the important feature that approximations to the derivatives of Lagrange multipliers are employed that avoid the need to calculate any second derivatives. Global convergence and local superlinear convergence results are proved. Numerical results are also given.




Cited in
(54)


Describes a project that uses

Uses Software






This page was built for publication: A recursive quadratic programming algorithm that uses differentiable exact penalty functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3734178)