Configuring the Mixed-Integer Sequential Quadratic Programming (MISQP) Technique

Mixed-Integer Sequential Quadratic Programming (MISQP) is a trust region–based method for solving problems that include integer and other discrete variables.

See Also
Mixed-Integer Sequential Quadratic Programming (MISQP) Technique
Configuring the Technique and the Execution Options
  1. From the Flow section of the action bar, click Optimization and drop it on the process diagram.
  2. Double-click Optimization .
    The Optimization Editor appears.
  3. From the General tab's Optimization Technique list, select MISQP.

  4. In the Optimization Technique Options area, enter or select the following:
    OptionDescription
    Max Iterations The maximum number of design iterations that you want the optimizer to run. The value type is integer. The default value is 40. Other possible values are ≥ 1.
    Max Branch and Bound Iterations Te maximum number of branch-and-bound iterations that you want the optimizer to run. The value type is integer. The default value is 40. Other possible values are ≥ 1.
    Termination Accuracy The termination criterion for MISQP. The MISQP stopping algorithm uses several alternative convergence checks, with the main convergence parameter based on the Karush-Kuhn-Tucker necessary optimality condition and the complementary slackness. The termination accuracy is applied in such a way that the scale of the objective and constraint parameters has little or no effect on the convergence check.

    The accuracy of the gradients calculation must be considered when selecting the Termination Accuracy value. If the subflow outputs are accurate up to 8–10 digits and the calculated gradients have at least 7 accurate digits, the recommended Termination Accuracy value is 1⁢ × ⁢10−7. If the gradient’s accuracy is lower, the Termination Accuracy value must be increased to 1.0⁢ × 10−5...1.0⁢ × 10−4. The value type is real. The default value is 1⁢ × ⁢10−6. Other possible values are > 0 and ≤ 0.1.

    Relative Step Size The relative finite difference step size for the creation of the linear process. The value type is real. The default value is 0.0010 (0.1%). Other possible values are > 0.0.
    Minimum Absolute Step Size The minimum absolute finite difference step for the creation of the linear process. The value type is real. The default value is 1⁢ × ⁢10−4. Other possible values are > 0.0.
    Use Central Differences Determine whether or not Optimization Process Composer will use the central difference method for calculating output derivatives. When this option is not selected, the forward difference method is used. Selecting this option increases the accuracy of the gradient calculations at the expense of doubling the number of design point evaluations.
    Maximum Failed Runs The maximum number of failed subflow evaluations that can be tolerated by the optimization technique. If the number of failed runs exceeds this value, the optimization adapter will terminate execution. To disable this feature, set this option to any negative value (for example, –1). When this option is set to a negative value, the optimization will continue execution despite any number of failed subflow runs.
    Failed Run Penalty Value The value of the Penalty parameter that is used for all failed subflow runs. The default value is 1⁢  ×  ⁢1030.
    Failed Run Objective Value The value of the Objective parameter that is used for all failed subflow runs. The default value is 1⁢  ×  ⁢1030.
  5. Click Ok to save your changes and to close the Optimization Editor.