Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Cantor's diagonal method

  1. Jul 13, 2005 #1


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    No question, or deep answers to be found here! I just wanted to share with you a pretty formulation of Cantor's diagonal argument that there is no bijection between a set X and its power set P(X). (the power set is the set of all subsets of X)

    It's based on the idea of a characteristic function: a function whose values are only 0 and 1. If Y is a subset of X, then I can describe Y by this function:

    [tex] \chi_Y(p) := \left\{
    0 \quad & p \notin Y \\
    1 \quad & p \in Y

    So that the statement "p is in Y" is equivalent to the equation "χY(p) = 1".

    With a little thought, you should be able to convince yourself that there is a one-to-one correspondence between subsets of X and characteristic functions on X.

    So, the question "Is there a bijection between X and P(X), the set of all subsets of X?" is equivalent to the question "Is there a bijection between X and the set of all characteristic functions on X?"

    Once you get to this point, the proof becomes much cleaner.

    We now assume that such a bijection exists, and we call it f. So, if x is an element of X, then f(x) is a characteristic function on X.

    If you haven't seen them before, function valued functions can be a little tricky to understand... for this proof, the only thing you need to know is that we can write f(x)(y), with both x and y in X, and f(x)(y) will either be a 0 or a 1.

    The next step is to write a function that represents the "diagonal": we define the function g by:

    g(x) := f(x)(x)

    And the key step is that we can now construct the function h:

    h(x) := 1 - g(x) = 1 - f(x)(x)

    The values of h are either 0 or 1, so it is a characteristic function on X. Now, since f is an invertible function (it's a bijection!), we can make this computation:

    h(f^{-1}(h)) = 1 - f(f^{-1}(h))(f^{-1}(h)) = 1 - h(f^{-1}(h))

    We can write f-1(h) because f is invertible, and the range of f is the set of all characteristic functions on X, and h is one of those things.

    The first step in the calculation is simply the definition of h. The second step uses the definition of invertibility, and will probably take a bit of thought to work through for those not comfortable with function valued functions. The key fact is this:

    [tex]f(f^{-1}(h)) = h[/tex]

    so if we apply both functions to an element of X, we have:

    [tex]f(f^{-1}(h))(x) = h(x)[/tex]

    Anyways, going back to the computation:

    [tex]h(f^{-1}(h)) = 1 - f(f^{-1}(h))(f^{-1}(h)) = 1 - h(f^{-1}(h))[/tex]

    This is a contradiction, because the far left and the far right can never be equal. (Remember that the values of h can only be a 0 or a 1, and this would require it to take 1/2 as a value)

    So, our initial assumption that there exists a bijection, f, from X to the set of all characteristic functions must have been incorrect, because we derived a contradiction. So, by the initial remarks, we have thus proven:

    There does not exist a bijection from any set X to P(X), the set of all subsets of X.
    Last edited: Jul 13, 2005
  2. jcsd
  3. Jul 13, 2005 #2


    User Avatar
    Science Advisor
    Homework Helper

    That's very pretty in a way :smile:

    Bit of an odd statement I know but I think so.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook