Zwischen Berechenbarkeit und Nichtberechenbarkeit. Die Thematisierung der Berechenbarkeit in der aktuellen Physik komplexer Systeme

Between Calculability and Non-Calculability. Issues of Calculability and Predictability in the Physics of Complex Systems. The ability to predict has been a very important qualifier of what constitutes scientific knowledge, ever since the successes of Babylonian and Greek astronomy. More recent is the general appreciation of the fact that in the presence of deterministic chaos, predictability is severely limited (the so-called ‘butterfly effect’): Nearby trajectories diverge during time evolution; small errors typically grow exponentially with time. The system obeys deterministic laws and still is unpredictable, seemingly a paradox for the traditional viewpoint of Laplacian determinisms. With the concept of deterministic chaos the epistemological issue about an adequate understanding of predictability is no longer just a mere philosophical topic. Physicists on the one hand recognize the limits of (long term) predictability, computability and even of scientific knowledge, on the other hand they work on concepts for extending the horizon of predictability. It is shown in this paper that physics of complex systems is useful to clarify the jungle of different meanings of the terms ‘predictability’ and ‘computability’ — also with philosophical implications for understanding science and nature. Today, from the physical point of view, the relevance of the concepts of predictability seems to be underestimated by philosophers as a mere methodological topic. In the paper I analyse the importance of predictability and computability in physics of complex systems. I show a way how to cope with problems of unpredictability and noncomputability. Nine different concepts of predictability and computability (i.e. open solution, sensitivity/chaos, redundancy/chance) are presented, compared and evaluated. This revised version was published online in August 2006 with corrections to the Cover Date.     

Naming and Diagonalization, from Cantor to Godel to Kleene

We trace self-reference phenomena to the possibility of namingfunctions by names that belong to the domain over which thefunctions are defined. A naming system is a structure of theform (D, type( ),{ }), where D is a non-empty set; for everya D, which is a name of a k-ary function, {a}: D^k D is thefunction named by a, and type(a) is the type of a, which tellsus if a is a name and, if it is, the arity of the named function.Under quite general conditions we get a fixed point theorem,whose special cases include the fixed point theorem underlyingGödel's proof, Kleene's recursion theorem and many othertheorems of this nature, including the solution to simultaneousfixed point equations. Partial functions are accommodated byincluding "undefined" values; we investigate different systemsarising out of different ways of dealing with them. Many-sortednaming systems are suggested as a natural approach to generalcomputatability with many data types over arbitrary structures.The first part of the paper is a historical reconstruction ofthe way Gödel probably derived his proof from Cantor'sdiagonalization, through the semantic version of Richard. Theincompleteness proof–including the fixed point construction–resultfrom a natural line of thought, thereby dispelling the appearanceof a "magic trick". The analysis goes on to show how Kleene'srecursion theorem is obtained along the same lines.