Cellular Automata Application on Chemical Computing Logic Circuits

Cellular Automata (CAs) have been proved to be a robust tool for mimicking a plethora of biological, physical and chemical systems. CAs can be used as an alternative to partial differential equations, in order to illustrate the evolution in time of the aforementioned systems. However, CAs are preferred due to their formulation simplicity and their ability to portray the emerging of complex dynamics. Their simplicity is attributed to the fact that they are composed by simple elementary components, whereas their complexity capacities are the result of emerging behaviors from the local interactions of these elementary components. Here, the utilization of CAs on mimicking of physio-chemical reactions is presented. In specific, the implementation of chemical-based logic circuits with the use of the Belousov-Zhabotinsky (BZ) class reactions was illustrated. The BZ reaction can demonstrate non-linear oscillations that have been utilized in different scenarios as a computational substrate, whereas its photo-sensitivity have been exploited as an additional factor of manipulating the computations. A common method to mathematically represent the BZ dynamics is the Oregonator equations, which are a set of PDEs. In this work the approximation of the Oregonator equations is performed with CAs to simulate logic circuits (from classic logic gates like AND to combinatorial ones). The proposed tool has been proved to be in agreement with results produced in the lab from the actual chemical reactions. Moreover, the tool is used to design novel computing architectures in a trivial manner, without the need of specialized knowledge on chemistry, without the need to handle dangerous chemicals and alleviating unnecessary costs for equipment and consumables. The main advantage of this method can be summarized as the acceleration achieved in current implementations (serial computers), but also towards potential future implementations in massively parallel computational systems (like Field-Programmable Gate Array hardware and mainly nano-neuromorphic circuits) that have been proved to be good substrates for accelerating the implemented CA models.