The tool supports interface of Foresight Analytics Platform 9 or earlier.
The panel is used to edit parameters of the criterion function of optimal control problem or criterion problem. The set of parameters on the panel depends on programming problem type.
Specify the following parameters of the criterion function:
Contents of the criterion function variables. Form a set of input variables (factors). Each variable is associated with an element of the Xn type, where n is the sequential number. These elements are used on creating the criterion function. Working with criterion function variables is similar to working with input variables.
To use variable in the criterion function:
Select a variable and click the Insert into Equation button.
Select the Insert into Equation item in the variable's context menu.
Drag the required variable from the list.
Enter identifier of the element corresponding to the variable.
Criterion function. Create the criterion function equation in the F= box. All elements corresponding to variables are shown as hyperlinks in the equation. Hovering over a hyperlink opens a tooltip containing the name of the corresponding variable. In the equation, you can use any arithmetic operations, parentheses, statistical, mathematical and custom functions or methods.
Click the Open in Editor button to edit the criterion function. The expression editor opens, in which enter the required changes.
After setting up the criterion function, select the method for optimizing the value of this function:
MIN. Minimum.
MAX. Maximum.
NOTE. An optimization option can be specified only for an optimal control problem and a criterion problem. An optimization option for a custom optimization problem can be set in a custom class that calculates the problem.
When the problem is saved, the criterion function correctness is checked. If the criterion problem formula contains errors (invalid operators, a non-existing element, and so on), an appropriate error message is displayed showing the number of the position containing this error. After the user closes the message, the cursor is moved to the indicated position of the formula.
Calculation method. The parameter is relevant only for the optimal control problem. Specify the method that is used to calculate the optimization problem:
Grid Search.
Sequential Quadratic Programming. This method is set by default.
Sequential Simplex.
Target trajectory. The parameter is relevant only for criterion problem. The target trajectory enables the user to set the required solution of criterion function at the forecasting period. To set target trajectory, click the Edit Target Trajectory button. The express report opens, in which enter the required values of target trajectory.
Accuracy of solution and number of iterations. The parameter is relevant only for criterion problem and optimal control problem.
Accuracy. Use the real number editor to determine the accuracy of problem calculations. The minimum and default value is 0.0001.
Number of Iterations. Use the value editor to define the number of iterations for problem calculations. The minimum value is 1; the default value is 1500.
Grid search | Sequential quadratic programming | Sequential simplex | |
Description | A grid search method. It refers to direct minimizing methods. It requires large calculation resources, and it is not designed to solve practical problems. | Method with quadratic subproblems. | Method of unconditional optimization. |
Parameters | It requires a defined range of valid values for each optimized variable. | It requires initial approximations to be set for all controlling variables. | It requires setting initial approximations. |
Constraints | It considers non-linear constraints of controlling variables. | It considers non-linear constraints of controlling variables. | It does not consider any constrains for the function arguments, by which optimization is executed. |
Derivatives | Calculating derivatives is not required. | Calculating partial derivatives is required. | Calculating derivatives is not required. |
Result | The result is the extremum of the criterion function found by grid search, and the number of cells is determined by the user. The accuracy of the found solution depends on the grid size. | The result is the local extremum determined by initial approximations or the extremum of the function when there are non-linear constraints on variables. | The result is the local extremum defined by initial approximations. |
See also: