Non-discrete biomolecular computing: an approach to computational complexity.

General principles of information processing at the molecular level inherent in simple biological and biomolecular entities can be used to elaborate essentially new non-discrete information-processing devices. These principles are: giant parallelism of information processing; processing mechanisms based on complicated non-linear dynamics; high efficiency of information transformations; considerable behavioral complexity of computational (pseudoelementary) primitives; and the possibility of variation and evolution of the molecular components of information-processing devices, including the possibility of evolutionary learning. Problems of high computational complexity are currently of great practical importance. Non-discrete biomolecular information-processing devices seem to be able to solve effectively some classes of problems of high computational complexity.