• Lvxferre@lemmy.ml
    link
    fedilink
    arrow-up
    9
    ·
    1 year ago

    A probability of 0.508 seems close enough to 0.5 for all practical matters. However if you need it to be exactly 0.5, you could instead flip the same coin twice, with the same side up in both starts. If you get two heads or two tails, disregard and repeat; else, your result is decided by tails-heads vs. heads-tails. Here’s the probability table for that:

    • p(same, same) = (0.508)*(0.508) = 0.258
    • p(same, opposite) = (0.508)*(0.492) = 0.250
    • p(opposite, same) = (0.492)*(0.508) = 0.250
    • p(opposite, opposite) = (0.492)*(0.492) = 0.242

    This also works with unfair coins. Let’s say that we got a coin with 0.7 likelihood to land on heads, 0.3 on tails:

    • p(H,H) = (0.7)*(0.7) = 0.49
    • p(H,T) = (0.7)*(0.3) = 0.21
    • p(T,H) = (0.3)*(0.7) = 0.21
    • p(T,T) = (0.3)*(0.3) = 0.09
    • mindbleach
      link
      fedilink
      arrow-up
      2
      ·
      1 year ago

      Sounds like the most comprehensible version of a whitening transformation.