lp_solve is a software package written in C that solves linear programs using a revised Simplex algorithm. It also includes branch-and-bound functionality to handle mixed-integer-linear programs. It's fairly portable, reasonably quick, and reasonably stable. I've run it on Sun Sparcs, Atari STs, and Linux systems.
PPRN Network Optimization Package The package is appropriate for solving a high variety of network problems: single/multicommodity network flow problems, with linear/nonlinear objective function and with/without linear side constraints. Thus it can be viewed as a general package for many network optimization problems (though it was originally designed for solving nonlinear multicommodity problems with linear side constraints). Distributed in binary form for Sun and DEC Alpha only.
The SLATEC Common Mathematical Library is an extensive public-domain FORTRAN source code library (consisting of more than 1400 routines and 300,000 lines of code and documentation!) developed and maintained by a consortium of Department of Energy and Department of Defense national laboratories. Version 4.0 of the library was released in December, 1992.
SLATEC incorporates within itself several other public domain packages, including EISPACK, FFTPACK, QUADPACK, FNLIB, BSPLINE, SLAP, and others.
There is a lot of material here. Datasets, documents, software, references to many sites associated with statistics.
Occasional snapshots of this archive are included in Prime Time Freeware for Unix, a semi-annual, mixed-media (book/CD-ROM) publication of Unix-related freeware.
Relax IV uses a relaxation-based network flow optimization algorithm created by Dimitri Bertsekas. This algorithm is supposed to be particularly good for larger network problems.
LOQO solves convex optimization problems to optimality and nonconvex ones to local optimality. While a subroutine library still exists, the simplest way to feed a nonlinear optimization problem to LOQO is via the math programming language AMPL . This is the default mode of operation. LOQO is free for academic use.
ASCEND IV is a free, large-scale, equation-based, environment featuring a strongly-typed, object-oriented model-description language. ASCEND is designed to reduce the time needed for creating, debugging, and solving mathematical models by orders of magnitude in comparison with C++-like and FORTRAN-like languages. Written using TCL 8.0, it runs on both Unix and WinTel systems.
Institute for Complex Engineered Systems/Engineering Design Research Center
Carnegie Mellon University
Written in Python , this includes a fairly sophisticated GUI front end.
Scilab is a high-level language for numerical computations in a user-friendly environment. Scilab is developed at INRIA and is available free via anonymous ftp in source and binary formats.
There exists a Matlab-to-Scilab translator, an interface to Maple, interfaces to scripting languages such as TCL, support for parallel processing using PVM, and many, many other libraries and toolkits.
Based on the LLVM compiler toolset, this wants to become a "better MATLAB," with decent capabilities at manipulating strings and files and such, on the general purpose computing side of things.
An "Open Source" optimization system (mostly in C++ ) including:
Branch-cut-price framework (BCP)
Cut generation library (CGL)
Open solver interface (OSI)
Which presently interfaces to OSL and XPRESS-MP
Fast approximation algorithms (FAA)
Derivative free optimization (DFO)
bonsaiG - a mixed-integer linear programming code
As a research code, bonsaiG is built more for flexibility and robustness than for speed. Compiled with full debugging capabilities (the default build), it's downright paranoid and can tell you far more than you'll ever want to know about what it's doing while it solves your MILP problem. Compiled with full optimisation, it's decently fast, but if you want real speed, buy a commercial code.
bonsaiG accepts problems in MPS format, and writes output in a form that's reasonably human-readable but conforms to no particular standard.
A set of routines written in C that
include a simplex method driver,
LP pre-solver, glpsol, a Java interface, and code to support solving
mixed integer and linear programs. It includes parsers for the common
MPS and CPLEX LP formats, as well as its own modelling
language, GLPK/L .
For pure linear programs, you can also use a primal/dual interior point solver. To use branch and bound to solve MILP problems, you must use the simplex method solver.