The emergence of data structures from local interactions

Abstract

This paper describes research on the use of explicitly programmable complex dynamical systems as a high-level parallel programming methodology. We show how problem solving can be seen as two processes interacting at a micro-level, namely selection and combination. These processes originate from the application of simple rules to pairs of elementary solution constituents.

The example discussed here is the emergence of trees. We show how to bias the system towards certain types of solutions. Finally, we discuss the conditions under which local minima problems can be avoided.