Introduction to artificial life. Incl. 1 CD-ROM.

*(English)*Zbl 0902.68198
New York, NY: Springer. xviii, 374 p. (1998).

This book is an introduction to the computational principles underlying the study of artificial life. After reviewing the theoretical foundations of artificial life, which lie in the field of automata theory (basically, von Neumann’s theory of self-reproducing, so-called cellular automata), the implications this view has on artificial life models are discussed in the framework of artificial chemistry (acting on artificial molecules, as simulated by virtual automata) and the idea of self-replicating code.

In order to understand the dynamics of living systems, the book then surveys major concepts underlying information theory (channels, coding, entropy, joint and conditional uncertainty, information, channel capacity, and information transmission capacity for genomes), as well as statistical mechanics and thermodynamics (phase space and statistical distribution function, ergodicity, thermodynamical equilibrium and relaxation time, energy, entropy, second law of thermodynamics, temperature, Gibbs distribution, nonequilibrium thermodynamics, first-order phase transitions). Armed with this battery of methods, the author then considers basic notions for describing the complexity of simple living systems, whether natural or artificial, as the stochastic transfer of information from an environment into the genome. This leads to the discussion of the Maxwell demon, Kolmogorov complexity, physical complexity, complexity of tRNA or in artificial life. Turning to the notion of self-organization of living systems, various theories of self-organized criticality are considered then. This concept allows one to take a look at the evolution of self-replicating systems from the viewpoint of statistical theories that admit critical points. Next, evolution is considered as a complex optimization process, with focus on geometrical and global properties of the underlying landscape without regard to the actual individuals evolving on it.

The fundamental notion of percolation is introduced by considering, e.g., site percolation, cluster size distribution in percolation systems, one-dimensional percolation systems, higher-dimensional Euclidean lattices in percolation theory, percolation on Bethe lattices, scaling theory, and the role percolation theory might have for theories of evolution. Then the question is raised whether natural fitness landscapes are fractal. The notion of fitness landscapes is introduced theoretically (as well as fractal landscapes, diffusive and nondiffusive processes, RNA landscapes), and it is discussed more concretely in terms of the “avida” world as a paradigm for landscapes created by the artificial chemistry of populations of self-replicating code. After that, some experiments related to speciation with the “avida” artificial life simulator are discussed, by exploring phenomena such as adaptation or species and genetic distance. Basic modes of information propagation and diffusion (information transport and equilibrium, information diffusion and waves) are discussed witin the context of “sanda”, a version of “avida” running on parallel supercomputers.

The next chapter treats adaptive learning at the error threshold, by considering copy mutations, cosmic-ray mutations. Eigen’s error threshold and molecular evolution as an Ising model. Finally, the book contains the user manual for the “avida” system and has attached to it a CD-ROM of the “avida” software, an experimental artificial life simulator. Simply put, “avida” is a population of self-replicating strings of computer code subject to random mutations, adapting to a complex information-rich landscape.

In order to understand the dynamics of living systems, the book then surveys major concepts underlying information theory (channels, coding, entropy, joint and conditional uncertainty, information, channel capacity, and information transmission capacity for genomes), as well as statistical mechanics and thermodynamics (phase space and statistical distribution function, ergodicity, thermodynamical equilibrium and relaxation time, energy, entropy, second law of thermodynamics, temperature, Gibbs distribution, nonequilibrium thermodynamics, first-order phase transitions). Armed with this battery of methods, the author then considers basic notions for describing the complexity of simple living systems, whether natural or artificial, as the stochastic transfer of information from an environment into the genome. This leads to the discussion of the Maxwell demon, Kolmogorov complexity, physical complexity, complexity of tRNA or in artificial life. Turning to the notion of self-organization of living systems, various theories of self-organized criticality are considered then. This concept allows one to take a look at the evolution of self-replicating systems from the viewpoint of statistical theories that admit critical points. Next, evolution is considered as a complex optimization process, with focus on geometrical and global properties of the underlying landscape without regard to the actual individuals evolving on it.

The fundamental notion of percolation is introduced by considering, e.g., site percolation, cluster size distribution in percolation systems, one-dimensional percolation systems, higher-dimensional Euclidean lattices in percolation theory, percolation on Bethe lattices, scaling theory, and the role percolation theory might have for theories of evolution. Then the question is raised whether natural fitness landscapes are fractal. The notion of fitness landscapes is introduced theoretically (as well as fractal landscapes, diffusive and nondiffusive processes, RNA landscapes), and it is discussed more concretely in terms of the “avida” world as a paradigm for landscapes created by the artificial chemistry of populations of self-replicating code. After that, some experiments related to speciation with the “avida” artificial life simulator are discussed, by exploring phenomena such as adaptation or species and genetic distance. Basic modes of information propagation and diffusion (information transport and equilibrium, information diffusion and waves) are discussed witin the context of “sanda”, a version of “avida” running on parallel supercomputers.

The next chapter treats adaptive learning at the error threshold, by considering copy mutations, cosmic-ray mutations. Eigen’s error threshold and molecular evolution as an Ising model. Finally, the book contains the user manual for the “avida” system and has attached to it a CD-ROM of the “avida” software, an experimental artificial life simulator. Simply put, “avida” is a population of self-replicating strings of computer code subject to random mutations, adapting to a complex information-rich landscape.

Reviewer: Udo Hahn (Freiburg i.Br.)

##### MSC:

68T99 | Artificial intelligence |

92D99 | Genetics and population dynamics |

82B99 | Equilibrium statistical mechanics |

68-01 | Introductory exposition (textbooks, tutorial papers, etc.) pertaining to computer science |