Learning Forum

[Sheng Lun Chou]

Hi ALL,

When I use the Maxwell Optimization function, I encounter the following problems. How can I solve them?

1. 1. Is it possible to control the minimum precision in optimization? For example, the optimization result of "para1" calculation is 0.0251125648, but the value after 3 decimal places cannot be controlled physically, so I hope it can only be calculated to 0.025.
2. Extending the first question, what is the definition of min step and max step in the variable? If I understand it literally, as shown below, I set them both to 0.02, and the minimum step should be 0.02, but why is the "0.385" value appears?
3. Can the calculated target values ​​of various calculation targets be directly viewed in View Analysis Result in the optimization results like Parametric result .
4. What is the meaning of Linear Constraint in Variable and initial disp. in Sensitivity, and how to use it?
5. As shown in the figure below, set "para2" to perform calculations on three different values ​​to find the best value of para1. Therefore, theoretically there will be three results, but there is only one calculation result. I would like to ask if any setting is wrong?

[GLUO]

Hi,

1. Maxwell should still use 0.0251125648 behind the scene. It shows 0.02 for easier display.

[GLUO]

2. In the Maxwell Optimization variable tab, the Min Step and Max Step are used to define the step size for the Quasi Newton and Pattern Search optimizers. The step size is the difference in a variable's value between one solved design variation and the next. This step size is determined when Optimetrics locates the next design variation that should be solved in an effort to meet the cost function.

[GLUO]

3. To view cost values versus completed iterations in data table format, you can right-click the optimization setup for which you want to view the cost results, and then click View Analysis Result on the shortcut menu. The Post Analysis Display dialog box appears. Under the Result tab, select Table as the view type. The cost value at each solved design variation is listed in table format.

[GLUO]

4. In Maxwell Optimization, the term "Linear Constraint" refers to the boundaries set for the solution of the minimization problem. In practice, this solution is sought only on a bounded subset of the Rn space, which is known as the feasible domain. This feasible domain is defined via linear constraints.

For initial disp and Q.5, please start a new post with more details, so more people can view and help.

Thank you,

GL