experience with Global Optimization application?
- To: mathgroup@smc.vnet.net
- Subject: [mg11229] experience with Global Optimization application?
- From: lars@gsblas.uchicago.edu (Lars A. Stole)
- Date: Mon, 2 Mar 1998 23:11:15 -0500
Has anyone had any experiences (good or bad) with the Global Optimization Application (a third-party application) discussed on Wolfram's website? More specfically, I've been using another system for doing nonlinear optimization, typically on problems with quadratic objective functions and a hundred linear constraints and choice variables. I don't expect Mathematica (epsecially with a grid search) to be as fast; I just want it to be tolerable (less than 30 minutes) as opposed to 30 seconds. Thanks in advance for any info. To quote from the Mathematicathematic website re "Global Optimization": Global Optimization is a package that performs nonlinear global optimization. It uses the Mathematica system as an interface for defining the nonlinear system to be solved and for computing numeric function values. Any function computable by Mathematica can be used as input, including degree of fit of a model against data and simulation models. The package uses a unique grid refinement algorithm. This algorithm is based on the identification of feasible points which define the solution set at each iteration. As lower points are found during the grid refinement process, points far from the current optimum are pruned from the solution set. As a result, multiple minima can be found in a single run--if they exist. The algorithm can also identify optimal regions rather than only a single point. These optimal regions might represent the bounds on feasible management strategies that achieve an equivalent result, or they might depict confidence limits for a parameter estimation problem. Nonlinear inequality constraints, which may even define disjunct parameter search spaces, are allowed. Advanced mathematics is not required to use Global Optimization. Derivatives are not required, and the function need not even be differentiable. Robust solutions are provided with the goal of saving time for the user by finding all optimal solutions in a single run. Developed and supported by Loehle Enterprises.