IBM Optimization Solutions and Library for Multi-platforms Version 2

IBM Optimization Solutions
DOES Miracles
for
Windows 95/NT, AIX, HP-UX,
Sun Solaris, SGI IRIX
 
Stochastic Extensions and Library on Linux (RedHat V6.2)
AMPL with OSL is available from OST
New news:Free Academic Licenses
  The Library and Extensions Products
The Solutions Products
LP, QP, MIP, and Stochastic
Reporting Problems, Getting Help
Download Software
 All the Products are Year 2000 Ready
How to get Production Licenses


Overview of the IBM Optimization
Solutions and Library

Mathematical optimization is concerned with the problem of finding an optimal allocation of limited resources by choosing an alternative that maximizes payoff, or minimizes cost, from among those that satisfy the given constraints. The IBM Optimization Solutions and Library family of products (hereinafter referred to as "OSL") comprises a suite of robust, high performance optimization applications (the IBM Optimization Solutions), and a library of functions (the IBM Optimization Library) from which a knowledgeable practitioner can create new, customized applications.

Optimization is applicable in many areas, including:

OSL is high quality optimization software implementing fast, robust, state-of-the-art algorithms. The Optimization Solutions products provide stand alone solver capabilities for analyzing linear programming (including network programming) problems, quadratic programming problems,  mixed integer programming problems, and stochastic programming problems. These Solutions incorporate modules from the Optimization Library, and provide most of the functionality of the complete library without requiring the user to develop code. For linear programming and quadratic programming problems, both simplex solvers and interior point solvers are included. The size of problems that the optimization solutions programs can handle is limited only by available memory.

The Optimization Library  includes about 70 user callable functions for manipulating models and analyzing the resulting mathematical problems. Besides the solvers, there are modules that analyze, simplify, and transform problems prior to solution, and others that assist with analyzing results. These post solution analysis capabilities include: infeasibility, sensitivity, and parametrics. The modules of the Optimization Library include many program options, control variables, and user exit function calls that provide great flexibility in creating specialized applications. The Optimization Solutions modules provide access to the program options and the control variables.

The Optimization Library Stochastic Extensions product works with the Optimization Library, bringing advanced stochastic programming algorithms together with the power and flexibility of the solvers and control capabilities of the Optimization Library. An innovative flexible nested decomposition algorithm allows the developer total control over partitioning of master problems into manageable sub-problems, thereby allowing solution of very large, multi-stage stochastic models. Quadratic risk variables are supported, and routines are provided for parametric analysis of solutions.

The Optimization Library Parallel Extensions product provides transparent parallelization of serial application programs that incorporate Optimization Library modules. More specifically, the Parallel Extensions product provides parallel versions of the Optimization Library interior point LP/QP solver, MIP solver, and LP decomposition module, and enables a user to specify whether, and on how many processors, these parallel solvers will run. The names (and calling sequences) of these parallel modules are precisely the same as the names (and calling sequences) of the corresponding serial modules. Furthermore the choice between serial and parallel execution can be made at run time, without altering or recompiling the source code.

OSL is available on numerous platforms, including: RISC System/6000Ò running AIXÒ , PCs running WindowsÒ (NT or 95), and UNIX based workstations from Hewlett Packard, Silicon Graphics, and Sun.

 Top of page


IBMhomeOrderPrivacyLegalContactIBM