Implementation in C

One would ask, why a C implementation of genetic programming is at all desirable. Koza does explain why he chose LISP, and the following lists reasons why C might be viewed as a bad tool to construct a genetic programming platform:

• C lacks the interpretive nature of LISP, and most certainly lacks the ease of representation.
• To write C code that generates other code and operates upon itself is a daunting task indeed.
• Another major drawback to the C approach is the lack of automated garbage collection and memory allocation. This can become a sore spot, since the members of the population change constantly, and, as will be expained shortly, makes the C implementation much more difficult.

• faster performance, both in speed and size. In general, the programs we wrote took up about one to two megabytes of memory. LISP interpreters can be very large, and because of the nature of the beast, they can be slow as well.
• greater portability. Most platforms support standard ANSI C code, and because of the relatively small memeory requirements, the C code can be compilied on a variety of machines with little or no change.
• better structure support. The flexibility of creating complex structure does open the possibility of some rather strange, and useful implementations.
• a familiar interface. C programming is commonplace, and therefore the implementation of genetic programming in C allows a wider audience to write and understand the code in which the underlying processes are written.

• The basic tree structure
• Generating an initial population
• Genetic programming made easy
• Results
• Further work