Programming Reaction-Diffusion Processors

In reaction-diffusion (RD) processors, both the data and the results of the computation are encoded as concentration profiles of the reagents. The computation is performed via the spreading and interaction of wave fronts. Most prototypes of RD computers are specialized to solve certain problems, they can not be, in general, re-programmed. In the paper, we try to show possible means of overcoming this drawback. We envisage an architecture and interface of programmable RD media capable of solving a wide range of problems. 1 Reaction-Diffusion Computers Reaction-diffusion (RD) chemical systems are well known for their unique ability to efficiently solve combinatorial problems with natural parallelism [2]. In RD processors, both the data and the results of the computation are encoded as concentration profiles of the reagents. The computation per se is performed via the spreading and interaction of wave fronts. The RD computers are parallel because the chemical medium’s micro-volumes update their states simultaneously, and molecules diffuse and react in parallel (see overviews in [1, 2, 8]). RD information processing in chemical media became a hot topic of not simply theoretical but also experimental investigations since implementation of basic operations of image processing using the light-sensitive Belousov-Zhabotinsky (BZ) reaction [28]. During the last decade a wide range of experimental and simulated prototypes of RD computing devices have been fabricated and applied to solve various problems of computer science, including – image processing [35, 3], – path planning [43, 12, 34, 6], – robot navigation [7, 10], – computational geometry [5], – logical gates [45, 39, 4], – counting [24], – memory units [30]. Despite promising preliminary results in RD computing, the field still remains art rather then science, most RD processors are produced on an ad hoc basis without structured top-down approaches, mathematical verification, rigorous J.-P. Banâtre et al. (Eds.): UPP 2004, LNCS 3566, pp. 31–44, 2005. c © Springer-Verlag Berlin Heidelberg 2005