[Home]
[
News]
[ Events]
[ People]
[ Research]
[ Education]
[Visitor Info]
[UCSD Only]
[Admin]
News]
[ Events]
[ People]
[ Research]
[ Education]
[Visitor Info]
[UCSD Only]
[Admin]
Directors:
Randolph E. Bank
Philip E. Gill
Michael Holst
Administrative Contact:
[ Click to Send Email ]
Randolph E. Bank
Philip E. Gill
Michael Holst
Administrative Contact:
[ Click to Send Email ]
Equations for a Projected-Search Interior Method for Nonlinear Optimization
by Philip E. Gill, Minxin Zhang
Abstract:
In CCoM Report 22-01, Gill and Zhang propose a primal-dual path-following method for general nonlinearly constrained optimization that combines a shifted primal-dual path-following method with a projected-search method for bound-constrained optimization. The method involves the computation of an approximate Newton direction for a primal-dual penalty-barrier function that incorporates shifts on both the primal and dual variables. This note concerns the formulation of approximate Newton equations for a nonlinear optimization problem in general form. These equations may be used in conjunction with a projected-search method to force convergence from an arbitrary starting point. It is shown that under certain conditions, the approximate Newton equations are equivalent to a regularized form of the conventional primal-dual path-following equations.
UCSD-CCoM-22-02.pdf February 2022
by Philip E. Gill, Minxin Zhang
Abstract:
In CCoM Report 22-01, Gill and Zhang propose a primal-dual path-following method for general nonlinearly constrained optimization that combines a shifted primal-dual path-following method with a projected-search method for bound-constrained optimization. The method involves the computation of an approximate Newton direction for a primal-dual penalty-barrier function that incorporates shifts on both the primal and dual variables. This note concerns the formulation of approximate Newton equations for a nonlinear optimization problem in general form. These equations may be used in conjunction with a projected-search method to force convergence from an arbitrary starting point. It is shown that under certain conditions, the approximate Newton equations are equivalent to a regularized form of the conventional primal-dual path-following equations.
UCSD-CCoM-22-02.pdf February 2022