In order to become familiar with the basic functions of MCMA we
suggest the following steps:
Figure 9:
Main menu of the MCMA application (with
the selection of the New problem item).
Figure 10:
Select a file dialog.
Figure 11:
Dialog for selecting a file with predefined criteria (with the main MCMA window in the background).
Figure 12:
Dialog for activating interactive selection of variables defining criteria.
- Select the New problem menu item
(as shown in Figure 913).
The list of available core models will be displayed
as illustrated in Figure 10.
Then select the file name corresponding to the core
model you want
to examine (in the current distribution the Windows'95/NT
version of the AEZ model is called aez_dos.cor)
and click the mouse on the OK button.
- Next, the dialog shown in Figure 11 will be
displayed.
This dialog allows for a specification of the name of a file containing
the predefined criteria
(see Section 4.2.1 for details).
The default file name is composed of the problem id (which
is the same as the root of the name of the file containing
the core model) and the string cri used
as the extension of the file name.
If the specified file is not readable, then another dialog gives
a choice between correcting the name of the file with
predefined
criteria and refraining from using any file with predefined criteria.
Figure 13:
Dialog for selecting a mask for names of variables.
Figure 14:
Dialog for selecting variables defining criteria.
- Next, the dialog shown in Figure 12 gives
an opportunity
of interactive selection of the core model
variables, which can be
used for specification of criteria
(such variables will be added to those used for
predefined criteria, if any).
If the Yes button is pressed for this dialog, then
the dialog titled Select a mask for a name
will be displayed (Figure 13).
This gives an opportunity to preselect a group of
the core model
variables with names that start with a given string (one or more
characters).
Typically a core model contains hundreds of variables
out of which only a few are used as criteria.
Selection of an empty string in this dialog would result in providing
all variables (which might result in displaying a dialog with
several thousands of variables for a large core model).
- In order to select only the variables that are used as
criteria in the tutorial,
type V as a mask name
and click the mouse on the OK button.
The window titled Variables defining criteria will be
displayed (Figure 14).
- Double click on the name V0000001 to copy this name
from the Selected variables list to the Criteria
defined by: list.
Do the same for names V0000002, V0000005, V0000008.
You may also click on the Help button to read more information
about this dialog.
- In order to add to the list of variables whose name
starts with another string click the mouse on the Add button.
The dialog titled Select a mask for a name will be displayed
(Figure 13) again
and you can add another group of variables (this is not required
for the described example).
Click the mouse on the OK button when all variables defining
the criteria will be selected.
- The next step is to define meaningful names and units for
the
criteria, to select the type for each criterion and/or to
select which predefined criteria will be used for the analysis.
This is done by the dialog titled
Definition of criteria
(Figure 15) which is displayed after the selection of
variables defining the criteria.
For the tutorial example we suggest to ignore all predefined
criteria except the following four: FoodAv, Land, NetRev and
TotEro.
The names of the criteria which were not predefined are set
to crit_? (where ? is replaced by a digit or
another character).
For such criteria meaningful names and units
should be defined.
Note that this is not necessary for criteria which were predefined.
If one does not use the predefined criteria,
we suggest to use the following names for interactively selected
variables: FoodAv for V0000001 criterion,
Land for V0000002, NetRev for V0000005,
and TotEro for V0000008.
Note that (due to the MPS format restrictions adopted by many
LP packages) the names of criteria are restricted to
6 characters (two more characters are needed by MCMA for creation
of unique names of the parametric optimization problem),
and should be composed of ASCII characters (without blank spaces).
The variable
column with the static strings contains the
names of core model variables that define the corresponding criteria.
Specify the criterion type maximize for the FoodAv and
NetRev criteria by clicking on the
corresponding radio buttons (the other two criteria will have
the default minimize type).
Double-check the criteria names and types and
click the mouse on the OK button when you are done.
Figure 15:
Dialog for definition of criteria (after suggested modifications).
Figure 16:
Dialog for definition of problem name.
Figure 17:
Information about computation of the pay-off table and a compromise solution.
- The next dialog (Figure 16)
will ask for the name of the problem.
You can define your name or you can use the predefined name by
pressing <enter> or by clicking on the OK button.
Completion of this dialog finishes the preparatory stage of
the interaction with
the user aimed at the definition of criteria.
Then the MCMA module starts a series of optimizations in order to compute
Utopia point, approximation of Nadir point,
and a compromise
Pareto-optimal solution.
For the n selected criteria this will require solving
(2n+1) LP problems each with about 600 variables and 300 rows.
The corresponding information will be displayed in a dialog
shown in Figure 17.
Please click OK and wait until the problem will be ready
for the interactive analysis.
You can follow the information about the generation of parametric
single-criterion optimization problems and update values of the
Utopia
point and approximation of the Nadir point which will be displayed
in the main MCMA window.
Once the computation of a compromise solution is completed
(which on a Pentium-90 PC took for this tutorial example14
about 3 minutes),
the information about the definition of criteria and the pay-off
table is stored by MCMA in a file that can be used later
for a continuation of analysis.
The computation of the compromise solution
is reported by a message shown in Figure 18,
and the default name of the main window of the MCMA will be changed
to the name you selected for the problem.
One can at this point store the compromise solution (by selecting
the Store item from the Solution menu) and exit
the program.
Figure 18:
Information about computation of the pay-off table and of a compromise solution.
Figure 19:
Main MCMA window after computation of the pay-off table and of a compromise solution.
- ...HREF="node26.html#fig:sap1">913
- MCMA run on other operating system will have a slightly different
appearance which is determined by a window manager used on such
a system.
However, the functionality of MCMA will be the same on
any operating system.
- ...example14
- This example is provided in the aez_test.cor file.
