Solution with LS-OPT

Solution with LS-OPTui

Solver

Define Solver and Input File

Select the Solvers tab.
For Solver Package Name select User-Defined because our example will not work with LS-DYNA but use Perl as the solver.
For Command specify perl (since this is the solver)
For Input File browse the file function.
For Name of Analysis Case enter 1.
Push the Add button.

Variables

Define the Variables

Select the Variables tab.
For Type of ”X” switch the combo box to Variable.
Enter -1 for the Minimum of the variable.
Enter 3 for the Maximum of the variable.
For Type of “Y“switch the combo box to Variable.
Enter -1 for the Minimum of the variable.
Enter 3 for the Maximum of the variable.

Sampling

Sampling Panel

Select the Sampling Panel.
Switch the radio button of the section “METAMODEL” to “Polynomial” and the radio button of the section “Order” to “Linear”. In “POINT SELECTION” should “D-Optimal” be selected and “The total number of Simulation Points” should be 5.
All these values should be there by default.

Responses

Responses Panel

Select the Responses panel.
Select UNSERDEFINED.
We obtain the output of the response with the shell command cat out.txt. This file was written by our perl script.
For the Response Name we enter "F".
Push the Add button to create the new response

Objective

Objective Function

Select the Objective panel.
Choose the Response Weight as objective

Constraints

Constraints.

Select the Constraints tab.
For Response “F” accept the default -inf. for Lower Bound and +inf for Upper Bound

Run

Run the Optimization

Select the Run panel.
For Number of iterations enter 10.
Push the Run button to start the optimization. (Save the project as com in the same folder where you'd put the file function)

Com-file

The created command file may look like this:

 "optimizacion"
Author "Fabiola"
$ Created on Tue May  5 10:01:57 2009
solvers 1
responses 1
$
$ NO HISTORIES ARE DEFINED
$
$
$ DESIGN VARIABLES
$
variables 2
 Variable 'X' 0
  Lower bound variable 'X' -1
  Upper bound variable 'X' 3
 Variable 'Y' 0
  Lower bound variable 'Y' -1
  Upper bound variable 'Y' 3

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$      OPTIMIZATION METHOD
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
Optimization Method SRSM

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$      SOLVER "1"
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$
$ DEFINITION OF SOLVER "1"
$
 solver own '1'
  solver command "perl"
  solver input file "/home/prak1/fabiola/B_2_variables/function"
$ ------ Pre-processor --------
$   NO PREPROCESSOR SPECIFIED
$ ------ Metamodeling ---------
  solver order linear
  solver experiment design dopt
   solver number experiments 5
$ ------ Job information ------
  solver concurrent jobs 1
$
$ RESPONSES FOR SOLVER "1"
$
 response 'F' 1 0 "cat out.txt"

$
$ OBJECTIVE FUNCTIONS
$
 objectives 1
 objective 'F' 1
$
$ CONSTRAINT DEFINITIONS
$
 constraints 1
 constraint 'F'
$
$ PARAMETERS FOR METAMODEL OPTIMIZATION
$
 Metamodel Optimization Strategy DOMAINREDUCTION
$
  iterate param design 0.01
  iterate param objective 0.01
  iterate param stoppingtype and
$                               
$ OPTIMIZATION ALGORITHM
$
 Optimization Algorithm lfop
$
$ JOB INFO
$
 iterate 10
STOP

Results

Optimizacion History For Response "F"

The Optimization History is a good tool to monitor the progress of our optimizacion. The view will be updated continuously after each iteration.

Start the LS-OPT Viewer (Select the View tab) and

Select Opt. History.
From Entity to Monitor select Response → F.

→ We can see, that with 3 iterations there is still no convergence.

→ At the 6'th iteration we can see that the response is getting close to the optimal point.

→ After 10 iterations a good convergence can be observed. We can conclude that perhaps less than 10 iterations would have sufficed to find the optimum.

Download

The complete data set (input and command files) is avaliable for dowloads as optimierung_2variables.tar.gz or optimierung_2variables.zip

Sections

Personal tools

Solution with LS-OPT