Propeller includes many kinds of genetic operators to create variation within the population. You can specify the rate of the variation genetic operators with the :variation map.

Variation

Propeller includes many kinds of genetic operators to create variation within the population. You can specify the rate of the variation genetic operators with the :variation map.

Some of the available genetic operators are described in this documentation. See the code for others and for details.

To add a new genetic operator you must add a case for the operator’s keyword in new-individual, calling existing or new utility functions that should be included earlier in this file.

Crossover

Crossover genetic operators take two plushy representations of Push programs and exchange genetic material to create a new plushy.

Function	Parameters	Description
crossover	plushy-a plushy-b	Crosses over two individuals using uniform crossover, one Push instruction at a time. Pads shorter one from the end of the list of instructions.
tail-aligned-crossover	plushy-a plushy-b	Crosses over two individuals using uniform crossover, one Push instruction at a time. Pads shorter one from the beginning of the list of instructions.

Addition, Deletion, Replacement, Flip

Addition, deletion, replacement, and flip genetic operators take a plushy and a rate of occurrence to create a new plushy.

Function	Parameters	Description
uniform-addition	plushy instructions umad-rate	Returns a plushy with new instructions possibly added before or after each existing instruction.
uniform-replacement	plushy instructions replacement-rate	Returns a plushy with new instructions possibly replacing existing instructions.
uniform-deletion	plushy umad-rate	Randomly deletes instructions from plushy at some rate.

Uniform Mutation by Addition and Deletion

Uniform Mutation by Addition and Deletion (UMAD) is a uniform mutation operator which first adds genes with some probability before or after every existing gene and then deletes random genes from the resulting genome. It has been found that UMAD, with relatively high rates of addition and deletion, results in significant increases in problem-solving performance on a range of program synthesis benchmark problems. When you run a problem in Propeller, you can specify the umad-rate to determine the frequency of addition and deletion.

:umad in the :variation map when running a problem will call uniform-addition and uniform-deletion with the umad-rate. Since uniform-addition and uniform-deletion are somewhat stochastic, you can use :rumad to ensure that the actual rates of addition and deletion are equal when mutating a genome.

New Individual

The function new-individual returns a new individual produced by selection and variation of individuals in the population based on the genetic operators provided in the :variation map.