Figure 37: Selection of problem for continuation of the analysis.
To continue the analysis one should (after starting the MCMA) replace the selection of New problem in the first step of the interaction described above by the selection of the Continue a session item from the Problem main menu item (also illustrated in Figure 9 and then by a selection of a file with the extension mc. After selecting the file aez_dos.mc the information shown in Figure 37 will be displayed.
If a continuation of analysis is selected, then the files with extension mc are shown in the dialog illustrated in Figure 37. Such files contain all information necessary for the continuation of the analysis. The information illustrated in Figure 38 is provided before the continuation of the analysis can be selected.
Figure 38: Information for continuation of analysis.
Figure 39: Initial iteration for continuation of analysis.
Figure 39 illustrates the first iteration screen for a continuation of the problem analysis. You may notice that the components achievement functions are set to be defined by Utopia and Nadir point.
We will now illustrate how to change the status of a criterion, which can be done by selecting the item Status from the main menu. One can make a criterion inactive or disregarded by selecting a corresponding radio box button. Selecting the original type of a criterion (defined in the initial stage to be either minimization, maximization or goal) makes a criterion active again. The dialog for changing criteria status is shown on Figure 40. For the sake of illustration we select the criterion Land to be inactive.
Figure 40: Changing status of criteria.
Figure 41: Preferences for sixth iteration.
Figure 42: Solution for sixth iteration.
Further on we will illustrate two attempts to select unrealistically high reservation levels for all criteria. Figure 41 shows a setting of reservation levels for all three active criteria to be close to the Utopia values. The corresponding solution is shown in Figure 42. This solution illustrates well one of the nice features of the ARBDS approach. Namely, instead of reporting an infeasible solution, (which would be the case, if reservation levels would be represented by hard constraints), a Pareto-optimal solution is found with the best possible criteria values which are all in this case worse than the corresponding reservation levels.
Figures 43 and 44 illustrate preferences and solutions, respectively for the seventh iteration, which was an attempt to improve the values of two criteria which had values below the reservation levels by a considerable relaxation of reservation level for the third criterion. The result of this attempt is negative, although the value of the FoodAv criterion has been improved, it is still worse than the reservation level, and the value of TotEro was slightly worse.
Figure 43: Preferences for seventh iteration.
Figure 44: Solution for seventh iteration.
Finally, Figures 45 and 46 show how one can achieve a very good value also for the TotEro criterion by relaxing the reservation level for the food production.
Figure 45: Preferences for eighth iteration.
Figure 46: Solution for eighth iteration.
| Janusz Granat | - Institute of Control and Computation Engineering |
|
| Marek Makowski | - International Institute for Applied Systems Analysis | |