Bridging Genetic Programming and Type Theory Research

Encoding domain constraints into fitness functions is a key concept in genetic programming. However, there are different moments in which the encoding can constrain the search process: at the language level, via grammargrammarconstriction; at the level of program generation, via a modified grammargrammarsampling; at the level of invalid individual discarding or fixing level; and at the fitness function level, incorporating constraint violations as fitness penalties. This chapter defines a general model for constraints in GP that supports all these moments and provides examples based on grammars and type systemstype systems. It also proposes combinations of constraints that have not yet been explored, such as Linear, Affine, or Session Types in genetic programming.