Models of molecular computing based on molecular reactions

• Authors: Grzegorz Rozenberg
• Venue: New Generation Computing
• Year: 2002

We discuss two models of molecular computing. The first one is based on an abstract formulation of two sorts of molecular reactions: enforcing and forbidden. The enforcing reactions are reactions that may happen, and are allowed to happen, in a given molecular system, while the forbidden reactions are detrimental for the system (e.g., leading to "incorrect" computations) and thus must be avoided. Hence computations in such a forbidding-enforcing system are driven by enforcing conditions (describing the enforcing reactions), but they are restrained by forbidding conditions (describing the forbidden reactions). The second model, called molecular landscapes, is geared towards the display of solutions. It consists of organisms (computing agents) functioning in a common environment which plays the role of a common communication medium for the organisms. When a molecular landscapes system works on a specific computational problem, each organism is working on this problem. But as soon as an organism M will get its solution to the problem, it modifies the environment, which from this moment on supports only the organisms that get the same solution as M. This is done through a selection mechanism that relies on selective competition which "kills" the losers. Since this selection mechanism interacts with the solution mechanism from the moment that computations are initiated, it can drastically increase the density of good solutions (molecules). In this way the molecular landscapes system achieves the goal of displaying solutions.