On the robust power of morphogenetic systems for time bounded computation
The time appears ripe to enrich the original idea of membrane computing with principles of self-assembly in space. To this effect, a first step was taken with the introduction of a new such family of models M systems (for morphogenetic system) that own a number of basic macro-properties exhibited by higher living organisms (such as self-assembly, cell division akin to mitosis and self-healing), while still only leveraging local interactions of simple atomic components and explicit geometric constraints of their constituting elements. Here we further demonstrate that, experimentally in silico, M systems are in general also capable of demonstrating these properties robustly after being assembled from scratch from some atomic components and entering a homeostatic regime. The results are obtained through a series of experiments carried out with an M system simulator designed to implement this kind of model by researchers interested in exploring new capabilities. We further define probabilistic complexity classes for M systems and we show that the model is theoretically capable of solving NP-complete problems in P-time, despite apparent problems of an implementation, such as kinetic and concentration bottlenecks.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Sosík, P., Smolka, V., Drastík, J., Bradík, J., & Garzon, M. (2018). On the robust power of morphogenetic systems for time bounded computation. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 10725 LNCS, 270-292. https://doi.org/10.1007/978-3-319-73359-3_18