7.E.5. Clustered optimization with the genetic algorithm
Clustering has many uses when a large number of computations can be done in parallel with each other. For
optimization routines, the genetic algorithm lends itself nicely to this case. The genetic algorithm creates large numbers
of populations from which the best simulation runs are used to generate an optimum design. Note that many
optimization routines do not operate like this. Many optimization routines need the previous simulation run to continue
the algorithms process. However, since the genetic algorithm generates a large number of simulations to be run for each
iteration, it clusters naturally.
• Open
working from a shared drive.
This is a tapering example that will be simulated via BPM. The goal is to modify the beginning and ending width of
each segment so that the output field is as close to the datafile target.fld as possible. Feel free to open target.pfd
to see a plot of the target field.
• Open the MOST dialog. Uncheck the Enable clustering box, enter single into the Output Prefix dialog,
and then press OK to run the optimization example. A number of generations of the genetic algorithm are
performed. (This example may take a few minutes). At the end of the optimization example run, the
MOST engine window should look like this:
No comments:
Post a Comment