Optimization is described in Oxford Dictionary of Computing as follows:

The process of finding the best solution to some problem, where "best" accords to predetermined criteria.¹

Traditional optimization techniques require problems that can be formulated as mathematical programming models of linear, nonlinear and integer types.² This is not the case in most large scale problems that arise in the industry. These problems are characterized by a set of common features that makes them extremely hard if not impossible to describe by mathematical programming models with advanced queuing theory and stochastic process theory. These common features are:

• Large solution space.
• Unknown characteristics of the solution space.
• Solution space shows discontinuities that do not allow for mathematical-numerical optimization of the task.
• Many soft restrictions for the solutions. Soft restrictions are defined as restrictions that do not necessarily have to be kept, but lead to worse results.

Fortunately there exists several context independent meta-heuristic procedures that are well suited for these large scale industry problems. Genetic Algorithms is a good example of one such optimization procedure. Several of these optimization algorithms are implemented in FACTS-Analyzer and used for solving optimization problems formulated with the help of simulation models built in FACTS-Analyzer, commonly referred to as simulation based optimization (SBO). More on how these types of SBO problems are formulated in FACTS-Analyzer can be found in the section Optimization setup.

Parallel and distributed

Implemented as a client-server system over the Internet, FACTS Analyzer is a parallel and distributed SBO software which supports multiple simulation experiments and SBO processes to run concurrently. Evaluations (either of an experiment or an optimization) are performed on a cluster of computers managed by a server. This architecture with parallel and distributed simulation and optimization will enable faster results to be gained. Beside faster results it will also mean client computers that need not be bothered and slowed down by heavy computing that might leave them more or less useless.

1. Ed. Soanes, C. & Stevenson, A. (2005): The Oxford Dictionary of English. Oxford University Press. Revised edition. ↑
2. Laguna, M. (1997): Optimization of Complex Systems with OptQuest. Graduate School of Business Administration University of Colorado. ↑

Copyright©  Evoma AB, 2022. All rights reserved.