Sophistication (complexity theory)

In algorithmic information theory, sophistication is a measure of complexity related to algorithmic entropy.

When K is the Kolmogorov complexity and c is a constant, the sophistication of x can be defined as^[1]

\operatorname {Soph} _{c}(x):=\inf\{\operatorname {K} (S):x\in S\land \operatorname {K} (x\mid S)\geq \log _{2}(|S|)-c\land |S|\in \mathbb {N} _{+}\}.

The constant c is called significance. The S variable ranges over finite sets.

Intuitively, sophistication measures the complexity of a set of which the object is a "generic" member.

See also

Logical depth

References

↑ Mota, Francisco; Aaronson, Scott; Antunes, Luís; Souto, André. "Sophistication as Randomness Deficiency" (PDF). doi:10.1007/978-3-642-39310-5_17.

Further reading

Koppel, Moshe (1995). Herken, Rolf, ed. "Structure". The Universal Turing Machine (2Nd Ed.). Springer-Verlag New York, Inc.: 403–419. ISBN 3-211-82637-8.
Antunes, Luís; Fortnow, Lance (August 30, 2007). "Sophistication Revisited" (PDF). doi:10.1007/s00224-007-9095-5.
Luís, Antunes; Bauwens, Bruno; Souto, André; Teixeira, Andreia (2013). "Sophistication vs Logical Depth". arXiv:1304.8046.

External links

The First Law of Complexodynamics

This article is issued from Wikipedia - version of the 5/13/2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.