• MBM@lemmings.world
    link
    fedilink
    English
    arrow-up
    1
    arrow-down
    2
    ·
    3 months ago

    Infinite seems like it’s low-balling it, then. 0% of problems can be solved by Turing machines (same way 0% of real numbers are integers)

    • Cows Look Like Maps@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      2
      ·
      3 months ago

      Infinite seems like it’s low-balling it

      Infinite by definition cannot be “low-balling”.

      0% of problems can be solved by Turing machines (same way 0% of real numbers are integers)

      This is incorrect. Any computable problem can be solved by a Turing machine. You can look at the Church-Turing thesis if you want to learn more.

      • MBM@lemmings.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        3 months ago

        Infinite by definition cannot be “low-balling”.

        I was being cheeky! It could’ve been that the set of non-Turing-computible problems had measure zero but still infinite cardinality. However there’s the much stronger result that the set of Turing-computible problems actually has measure zero (for which I used 0% and the integer:reals thing as shorthands because I didn’t want to talk measure theory on Lemmy). This is so weird, I never got downvoted for this stuff on Reddit.

    • DaleGribble88@programming.dev
      link
      fedilink
      English
      arrow-up
      1
      ·
      3 months ago

      The subset of integers in the set of reals is non-zero. Sure, I guess you could represent it as arbitrarily small small as a ratio, but it has zero as an asymptote, not as an equivalent value.

      • MBM@lemmings.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        3 months ago

        The cardinality is obviously non-zero but it has measure zero. Probability is about measures.