To solve an optimization problem, right-click in the model tree and select compile and solve model. At best, the model succeed optimization to solve the problem. However, it may happen that there is no solution, perhaps because the constraints you have entered is too harsh. There are a variety of measures to identify where the bottlenecks are, by reformulating the model and loosen restrictions. This tool requires relatively high competence in modeling optimisation problems (but it is always possible to get help!)

 

 

OptModel_ContextMenu_Solve

Figure 1.

 

 

<%EXTOGGLE%>Save the optimisation results

If a solution is found, the program prompt you to save the optimization results. After that, you can use the report generator to create tables and graphs of the result, and view the results in the result map viewer. If you decline to save the results, you can do it later (but before you run another optimization) by selecting Save optimisation result button. Note the difference from Save model (third button from left), which saves the model structure but not the data and results.

 

 

OptModel_toolStrip

Figure 2.

 

<%EXTOGGLE%>Using penalty variables to hanle infeasible solutions

If the solver doesn’t find a solution for a problem, you can add penalty variables to the restrictions that you suspect are the cause of the infeasibility. The idea is to add one or more temporary penalty variables (in some literature these are called “artificial slack variables”). These are defined to compensate for shortages, or subtract unwanted surpluses. You sum up the values of the penalty variables and set a negative weight (penalty) on this sum in the target function (if this is a maximization problem, in a minimization problem you would use a positive weight). Then you can inspect the values of the penalty variables to identify which restrictions that cause the problem.

 

See also: HeurekaWiki: Optimisation settings

Contact support. © Sveriges Lantbruksuniversitet