Object Oriented Population Generation

Computational population generation produces individual records using Monte-Carlo techniques. Generation rules for each individual characteristic potentially depend on other characteristics which create a dependency tree for computing each individual. In addition evolutionary computation objectives define selection criteria of individuals to match aggregate population goals. These elements repeat many times when creating new populations, therefore code reusability is important to simplify data entry. Moreover it allows exchanging computational building blocks to create simulation variations for sensitivity analysis or hypothesis testing which are essential for reference modeling. This paper describes an object oriented approach where population building blocks can be inherited to build a population that later can be rebuilt with variations. The paper describes the modular modeling theory with some basic examples.