Soft Computing Evaluation Logic

3.9System Optimization

System optimization can be defined as a problem of finding the best configuration of an evaluated system, according to some optimization criterion. The optimization criterion is usually an appropriate combination of low cost and high suitability of the analyzed system. In an ideal case of system comparison and selection, we should optimize the configuration of each of competitive systems before the cost/suitability analysis and system selection.

3.9.1 Three Fundamental Constrained Optimization Problems

In a general case, a complex system has the total cost C and satisfies user requirements at the level of overall suitability S. In some cases, there are many possible configurations defined using various values of attributes. Typical examples of such decision problems are encountered in the area of computer evaluation and selection. While decision problems are human problems and they don’t change over time, both computer hardware and software change rapidly every year, and the latest shiny computer becomes obsolete before a book is printed. So, if we want to study computer optimization problems (because of their complexity), then we can only use obsolete machines. For example, Table 3.9.1 shows the decision problem facing a buyer of a typical 2004 legacy model of a desktop PC. By combining various independent components of hardware and software it is possible to make K = 2.92°10¹⁶ different configurations; one of them is the most suitable, and we want ...