• anton@lemmy.blahaj.zone
    link
    fedilink
    arrow-up
    4
    ·
    2 days ago

    Once you accept that each float represent a range of reals most design decisions make sense.
    Infinity is just the name given to values that are too large to represent otherwise and negative zero includes negative values too close to zero.

    What mathematics probably hate most about floats is that rounding breaks commutativity.

    • bss03@infosec.pub
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 day ago

      What mathematics probably hate most about floats is that rounding breaks commutativity.

      Actually, the maths people in my circles tend to complain more about how the normal floating-point equality test doesn’t define an equivalence relation, but really any of the violations of field properties can attract complaints.

      • anton@lemmy.blahaj.zone
        link
        fedilink
        arrow-up
        2
        ·
        edit-2
        20 hours ago

        When writing my comment I had successfully repressed that fact in my mind. That’s obviously worse.