Procedure for executing optimization

This page demonstrates how to create a program for conducting optimal design using pyfemtet.opt in your own project.

Creating a Femtet project

Create an analysis model on Femtet. Register the parameters you want to optimize as variables. For more details on analysis settings using parameters, please refer to Femtet Help / Project Creation / Variables.

Setting the objective function

In optimization problems, the metric to be evaluated is referred to as the objective function. Please write the process of calculating the objective function from analysis results or model shapes using Python macros in Femtet.
"""Example to calculate max displacement (for your obejctive function).
The scripts after Dispatch are Femtet's Python macros.
"""
from win32com.client import Dispatch

# Get object to control Femtet.
Femtet = Dispatch("FemtetMacro.Femtet")

# Open analysis result by Femtet.
Femtet.OpenCurrentResult(True)
Gogh = Femtet.Gogh

# ex.) Get max displacement from analysis deresult.
dx, dy, dz = Gogh.Galileo.GetMaxDisplacement()
Note

For the Python macro syntax in Femtet, please refer to the Femtet Macro Help or Macro Examples.

Creating the main script

Using the design variables and objective function defined above, create the main script.
"""The minimum code example to execute parameter optimization using PyFemtet."""

from pyfemtet.opt import FEMOpt

def max_displacement(Femtet):
    """Objective function"""
    Gogh = Femtet.Gogh
    dx, dy, dz = Gogh.Galileo.GetMaxDisplacement()
    return dy

if __name__ == '__main__':
    # prepareing optimization object
    femopt = FEMOpt()

    # parameter setting
    femopt.add_parameter('w', 10, 2, 20)
    femopt.add_parameter('d', 10, 2, 20)

    # objective setting
    femopt.add_objective(max_displacement, direction=0)

    # run optimization
    femopt.optimize()
Note

For this script to actually work, you need a Femtet stress analysis project with variables w and d.

Note

The objective function must take a Femtet instance as the first argument, since the FEMOpt instance intarcreates it internally.

Warning

Only perform add_parameter() on variables set with constant expressions within Femtet, and do not do it for variables set with string expressions. The string expressions will be lost.

Run the script.

When the script is executed, the progress and results will be saved in a csv file. Each row in the csv file represents the result of one analysis attempt. The meaning of each column is as follows:

Columns

Meaning

trial

The number of the attempt

<Variable name>

The value of the variable specified in the script

<Objective name>

The calculation result of the objective function specified in the script

<Objective name>_direction

The target of the objective function specified in the script

<Constraint name>

The calculation result of the constraint function specified in the script

<Constraint name>_lb

The lower bound of the constraint function specified in the script

<Constraint name>_ub

The upper bound of the constraint function specified in the script

feasible

Whether the attempt satisfies all constraints

hypervolume

The hypervolume up to that attempt (only when the objective function is 2 or more)

message

Special notes from the optimization process

time

Time when the attempt was completed

Note

Items enclosed in <> indicate that their content and number may vary depending on the script.

Note

If the objective name and constraint name are not specified in the script, values such as obj_1, cns_1 will be automatically assigned.