InformationTheory

InformationTheory.Shannon.Kolmogorov.Noncomputable

source

Kolmogorov complexity is not computable #

complexity grows unboundedly (exists_incompressible) yet, were it computable, one could search for the least string whose complexity reaches a given bound k. That search is a partial recursive description of an incompressible string using only the natLen k bits of k, so the invariance theorem bounds its complexity by natLen k + O(1). Since natLen is the binary length and grows logarithmically, the two facts collide: the searched string has complexity at least k but at most natLen k + O(1) < k for large k (Berry's paradox).

Main statements #

theorem

InformationTheory.Kolmogorov.exists_condIncompressible

source
(y k : ) :
∃ (x : ), k condComplexity x y
Used by
    theorem

    InformationTheory.Kolmogorov.condComplexity_not_computable

    source
    (y : ) :

    Conditional Kolmogorov complexity is not computable, for any fixed condition y (Berry's paradox): a computable C(· | y) would let one search for the least string of complexity at least k, describing it in only natLen k bits. @audit:ok

    Used by
      theorem

      InformationTheory.Kolmogorov.complexity_not_computable

      source

      Kolmogorov complexity is not computable (Berry's paradox). @audit:ok

      Used by