Optimization Software Developed by J. Nocedal's Research Team

Five optimization packages, KNITRO, L-BFGS, PREQN, CG+ and Wedge, are available.

L-BFGS, PREQN, CG+, and Wedge, can be freely used for research, education or commercial purposes,.
KNITRO is available for academic use for a nominal fee. Commercial licenses for KNITRO are available
(see below).

Disclaimer
This software is provided without any expressed or implied warranty. In particular, there is no warranty of any kind concerning the fitness of this software for any particular purpose.

KNITRO: General Nonlinear Programming Solver

KNITRO, a package for solving unconstrained and constrained optimization problems. Commercial, academic and government licenses are available.

Download a free student/evaluation edition of KNITRO here.

L-BFGS: Limited Memory Codes

Two codes for large-scale optimization are part of this package:

L-BFGS is a code for solving unconstrained problems. Author: J. Nocedal
L-BFGS-B is capable of solving problems with simple bounds on the variables. Authors: C.Zhu, R. Byrd, and J. Nocedal

For more information about the limited memory method see the NEOS Optimization Guide. Numerical results comparing L-BFGS and MINOS can be found here

PREQN: Software for Preconditioning the Conjugate Gradient Method

The package PREQN automatically generates preconditioners when solving a sequence of linear systems of equations by means of the conjugate gradient method. Authors: J.L. Morales and J. Nocedal

CG+: Conjugate Gradient Software

The package CG+ is a Conjugate Gradient code for solving large-scale, unconstrained, nonlinear optimization problems. CG+ implements three different versions of the Conjugate Gradient method: the Fletcher-Reeves method, the Polak-Ribiere method, and the positive Polak-Ribiere method (Beta always non-negative). Authors: G. Liu, J. Nocedal and R. Waltz

Wedge: Derivative-Free Optimization

WEDGE is a Matlab code for solving unconstrained optimization problems in which
the objective function is smooth and the number of variables is moderate, but derivatives are not available. Author: M. Marazzi.

Email Jorge Nocedal: nocedal@ece.nwu.edu