数学建模优化模型OptimizationModelling11.pdfVIP

11 Output Analysis and Practical Issues 11.1 Introduction After solving a mathematical model, the next stages of the decision process (as of Figure 2.1) are to perform a sensitivity analysis and then implement the solutions. Before implementation, however, the solutions must be analyzed and interpreted in a nontechnical manner and in nonmathematical terms so that nonexpert users can understand, accept, and implement the solutions. Otherwise, the whole effort would be to no avail, which frequently occurs in practice. In this chapter, we will discuss the solutions of models and the outputs provided by different software packages, and some practical issues faced by the users when dealing with optimization models and software packages. A number of examples to demonstrate these issues will be provided. 11.2 Solutions and Reports When model solving, we usually expect to find the values of the decision variables and the objective function. Depending on the model type, it is possible to generate much more information, which may assist in arriving at better decision alternatives. For example, information derived from sensitivity analysis and parametric programming in linear programming (LP). In addition to computationally more efficient algorithms, the ability of the algorithm to perform and report upon associated sensitivity analysis has made LP the most popular optimization tool. However, due to the discrete nature of the solution space, sensitivity analysis is not automatic for integer programming models and limited sensitivity information can be generated for nonlinear models. In solving any LP model, Excel Solver provides three report options: Answer, Sensitivity, and Limits. Any or all of these reports can be saved after a model has been solved. LINDO=LINGO and MPL=Conopt 2007 by Taylor Francis Group, LLC. Microsoft Excel 11.0 Answer Report