Peter Dittrich, Jens Ziegler and Wolfgang Banzhaf

University of Dortmund, Dempt. of Computer Science, Chair of Systemsanalysis (LS XI)

Mesoscopic Analysis of Self-Evolution in an Artificial Chemistry


In an algorithmic artificial chemistry the objects (molecules) are data structures and the interactions (reactions) among them are defined by an algorithm. The same object can appear in two forms:

Thus, the same object can act on other objects or it can be processed by others. This dualism allows the implicit definition of constructive artificial chemical systems, which exhibit quite complex behaviors. In our case even evolutionary behavior has been observed which is notable, because no explicit mutation, recombination or fitness function is involved. Every variation is exclusively performed by the objects (molecules) in their machine form. In addition to microscopic methods (e.g. monitoring the actions of single molecules) and macroscopic measurements (e.g. diversity, complexity) we developed a stepwise mesoscopic analysis method based on classification and dynamic clustering. Knowledge about the system is accumulated in a cyclic process, where measuring tools (classifiers) extract information, which is again used to create new classifiers.


self-organization, strong artificial life, evolution, computational chemistry, visualization, constructive dynamical systems, algorithmic chemistry, chemical computing, cluster analysis, binary string system, automata reaction

full paper

"Mar 13 2019" Peter Dittrich