The NEOS Server offers PGAPack (PGAPack Parallel Genetic Algorithm Library) for the solution of simple bound constrained optimization problems in AMPL format.
PGAPack was developed by David Levine of the Mathematics and Computer Science Division at Argonne National Laboratory.
The PGAPack AMPL interface was developed by Ismael Vaz in order to benchmark the PSwarm solver with a set of problems.
PGAPack makes no use of derivative information of the objective function (so modest problem size should be submitted).
For further information about PGAPack interface to AMPL,
For further information about PGAPack contact the PGAPack author.
Using the NEOS Server for PGAPack
The user may submit a model in
Examples of models in AMPL format can be found in the
The model is specified by a model file, and optionally,
a data file and a commands file.
If the command file is specified it must contain
the AMPL solve command.
The commands file can contain any AMPL command
options for ASA with, for example,
option pgapack_options "maxfun=1000";
Options include any of the parameters
- crossover - PGASetCrossoverType (1=ONEPT, 2=TWOPT, 3=INIFORM)
- debug_level - PGASetDebugLevel (12=ENTERED, 13=EXIT, 80=MALLOC, 82=PRINTVAR, 22=SEND, 23=RECV)
- fitness - PGASetFitnessType (integer)
- ga_iter_value - PGASetMaxGAIterValue (integer)
- maxfun - Maximum number of objective function calls (integer)
- mutation - PGASetMutationType (1=RAW, 2=NORMAL, 3=RANKING)
- num_replace_value - PGASetNumReplaceValue (integer)
- pop_rep - PGASetPopReplaceType (1=BEST, 2=NOREP, 3=REP)
- pop_size - PGASetPopSize (integer)
- selection - PGASetSelectType (1=PROPOR, 2=SUS, 3=TOUR, 4=PTOU)
Printing directed to standard out is returned
to the user with the output.