Modeling and optimization

Calculation logic principles

You can build a mathematical model by adding sequences, variables, parameters, objective functions, and constraints. Then, you can configure the solving properties and call the underlying solver to solve the model.

Procedure

1. Add a data source.

   Import the DB-IN component. Configure the database and its fields. Select the data required for the model.

2. Define the problem type.

   Select linear programming, non-linear programming, or constraint-based problems.

3. Add a model sequence.

   This creates an index. When you create the index, set Index Type to Normal Index and specify an Index Name. This name is used as a reference in model building. You can obtain the Index Value by manual input or by associating a table.

4. Define variables.

   During variable definition, if a variable has a subscript, add an index for it. You must also specify the variable's Feasible Domain and Initial Value.

5. Add parameters.

   Adding parameters is similar to adding variables. You can obtain the values by manual input or by associating a table.

6. Add objective functions and constraints.

7. Define solving parameters.

   Set the solving time to limit the computation duration. If an optimal solution is not found before the maximum runtime is reached, the process automatically exits and returns the best solution found so far.

8. Configure optimization data outputs.

   When you configure the outputs, you can choose whether to write the data back. If you do, select the target database and specify the write rules.

Parameter descriptions

• IN1 port - Input parameters

  The input port supports methods such as IGATE-IN and DB-IN.

• OUT1 port - Outputs

  Field	Value type	Description
  model_id	string	The model run ID (ProjectID_CanvasID_YYYYMMDDHHMMSS).
  model	string	The JSON information of the model content.
  start_time	datetime	The start time of the solving process.
  end_time	datetime	The end time of the solving process.
  computation_time	float	The runtime in seconds.
  obj	double	The result of the objective function.
  solver_info	string	The information about the solving result.
  variables	string	The JSON information of the variable list: { "var1_key1_key2_key3": { "value": 0.0, "keyName": "var1_key1_key2_key3", "quality": "-1" } }