Title :
Increasing the Gap between Descriptional Complexity and Algorithmic Probability
Author_Institution :
Sch. of Math., Stat. & Oper. Res., Victoria Univ. of Wellington, Wellington, New Zealand
Abstract :
In this paper we seek to analyze the relationship between two different ways of understanding the intrinsic algorithmic randomness of strings and real numbers. In the setting of discrete spaces, a cornerstone of the theory of prefix-free Kolmogorov complexity is the coding theorem. This theorem states that the negative logarithm of the probability that a string is the output of a universal prefix-free machine coincides, within some additive constant, with the length of its shortest description. It has been suggested that this ties two fundamental principles together: Bayes´ theorem and Occam´s razor. Certainly, the result lends support to the belief that prefix-free complexity is a natural measure of the algorithmic randomness of strings.
Keywords :
computational complexity; probability; Bayes theorem; algorithmic probability; coding theorem; descriptional complexity; discrete spaces; intrinsic algorithmic randomness; negative logarithm; prefix-free Kolmogorov complexity; universal prefix-free machine; Algorithm design and analysis; Codes; Computational complexity; Information theory; Length measurement; Mathematics; Operations research; Probability; Statistical analysis; Topology; Algorithmic randomness; Kolmogorov complexity;
Conference_Titel :
Computational Complexity, 2009. CCC '09. 24th Annual IEEE Conference on
Conference_Location :
Paris
Print_ISBN :
978-0-7695-3717-7
DOI :
10.1109/CCC.2009.13