introducing initial population to MOGA by DAKOTA

[Mr. hamid reza 1]

Hi 

I am using the Dakota as the optimizer engine. Now, I need to introduce an initial population to the MOGA. It is possible via "flat_file" option for the "initialization_type", in the Dakota input.
As you my see in the Dakota email lists, the Dakota has removed the related bug to the mentioned option in the version 6.7, which is recently released. I replaced the content of Dakota tool ( "C:\Program Files (x86)\FRIENDSHIP-SYSTEMS\CAESES\tools\dakota\6.3.0" ) with the content of 6.7.0 version.
I also, created the “initial_population.dat” in the folder of Dakota input (“C:\Program Files (x86)\FRIENDSHIP-SYSTEMS\CAESES\etc\dakota”).

But still when I run the optimization with the related Dakota input, the process is beginning with the random initial population and do not use the “flat_file” inputs and I have the following error in the Dakota ".log" file, which means the the Dakota is not able to read the data:

“quiet- flat_file: Encountered fatal error while attempting to read file "initial_population.dat".  Make sure the file exists and is a JEGA Design flat file.”

 

Is there any idea to help?

Best Regards
Hamidreza

• Reply
• Report
•  
•  

[Mr. Heinrich von Zadow 9]

Dear Hamidreza,

please use the following custom template. As for the *.dat file, you need to format it in komma-seperated columns for the design variables followed by the evaluations. The Location of the *.dat file on your computer needs to be specified in the template in line 11.

 

If you leave the columns for the evaluations empty, dakota will create the same designs, but run the computations and evaluations again.

 

Best regards,

Heinrich

#name Global Optimization

#docu 
Global optimization using the genetic algorithm MOGA. It can be used for global optimization tasks. Note that, in general, genetic algorithms need many evaluations. This makes this method suitable only for problems where the evaluation is not too expensive.
#!docu

method
  moga
  output debug
# set absolute path to existing resultPool.dat file:
  initialization_type flat_file = '\resultPool.dat'
  crossover_type shuffle_random
  num_offspring = 2 num_parents = 2
  crossover_rate = 0.8
  mutation_type replace_uniform
  mutation_rate = 0.1
  fitness_type domination_count
  replacement_type below_limit = 6
  shrinkage_percentage = 0.9
  niching_type distance 0.15
  postprocessor_type
  orthogonal_distance 0.05
  convergence_type metric_tracker
  percent_change = 0.05 num_generations = 10
  <max_function_evaluations, unsigned, 300, Max. Evaluations, set maximum number of function evaluations>
  <population_size, unsigned, 100, Initial Population Size, set size of initial population which might change during the run>
  <max_iterations, unsigned, 10, Max. Generations, set number of maximum generations>
 
variables
  continuous_design = [dv_count]
  initial_point [dv_initial_values]
  lower_bounds [dv_lower_bounds]
  upper_bounds [dv_upper_bounds]
  descriptors [dv_names]

responses
  objective_functions = [eval_count_objective]
  nonlinear_inequality_constraints = [constr_count_inequality]
  nonlinear_equality_constraints = [constr_count_equality]
  no_gradients
  no_hessians
  
[interface]
  asynchronous
  <evaluation_concurrency, unsigned, 8, Parallel Evaluation, number of concurrent designs, General, 3, false>

[Jörg 29]

Note that the next CAESES version will include this mechanism so that users do not have to take care of it.