Search: Lycos Angelfire     Texas Hold 'Em
share this page Share This Page  report abuse Report Abuse  build a page Edit your Site  show site directory Browse Sites  hosted by angelfire
    « Previous | Top 100 | Next » hosted by angelfire

Null set uniqueness theorem

Paul Conant's home page
When is truth vacuous? Is infinity a bunch of nothing?
John Allen Paulos. Probability and politics
D.J. Velleman. Set theory and logic
Herbert B. Enderton. Set theory and logic

Comment: Some are puzzled by the standard set theoretic fact that even when set A does not equal set B, A - A equals B - B (also written A\A and B\B). To beginners, it is sometimes counterintuitive that the empty subset of A is indistinct from the empty subset of B.


We accept these rules of inference, letting "~" stand for "not":

   The expression
i)    x ® y
      means "x implies y", which is equivalent to

ii)   if x is true, then y is true 
      which is equivalent to

iii)  ~x or y
      which "means either x is false, or y is true".

Example:

i)    A person's being alive implies that he or she has
      a beating heart

ii)   If a person is alive, then he or she has a
      beating heart

iii)  Either a person has a beating heart, or
      he or she is dead     
This permits us to write "x ® y" as "~x or y" (although English idiom often has "y or ~x" but the statements are equivalent as a truth table check shows).

We take "x « y" to mean "x is true only if y is true and y is true only if x is true" which can be written (x ® y) and (y ® x)


We accept these definitions and conditions a priori:

A set is defined by its elements, not by its descriptions. For example, let A = {r|r is a real root of x2 -5x + 6 = 0} and B = {2,3}. In that case, A = B.

Every element is unique. That is, if A = {2} and B = {2,2}, then A = B.

A set is permitted to be an element of another set, (but in standard theory it can't be a member of itself). By this we see that every set is unique.

A subset of a set is itself a set.

B subset of A means (using "e" for "element of"):

x e B ® x e A

Equality means

x e B « x e A, meaning B is a subset of A and A is a subset of B.


First, let us see that every set has a an empty subset, meaning you cannot remove the null subset from a set.

Let us provisionally call B a subset of A and then require that B have no elements.

By definition, x e B ® x e A, which is equivalent to

x ~e B or x e A

which is true, since x is not a member of B. Even if x is not a member of A (perhaps A is empty), the statement remains true.

That is, a subset with no elements satisfies the definition of subset.

Now suppose the statement that our belief that any set A has a null subset is false. Using { }A to mean null subset of A, we have:

~({ }A subset A), or,

~(x e { }A ® x e A), or,

~x ~e { }A or x e A), or,

x e { }A and x e A,

a contradiction, even if A is also empty.

Hence our suggestion is correct.

Now to prove uniqueness we simply need show that { }A = { }B for arbitrary sets A and B. We have

x e { }A « x e { }B, or,

(x ~e { }A or x e { }B) and (x ~e { }B or x e { }A).

To the left of the "and" we have the truthfulness of x not being an element of { }Aand similarly to the right.

Hence { }A = { }B

Remembering that any subset is a set, our claim that only one null set exists is verified.


It is also instructive to verify the uniqueness of the null set by considering complement sets.

Adopting the notation for the complement set, we have

x e A\B ® x ~e B

We know that A = A and so A is a subset of A. This fact allows us to write:

x e A\A ® x ~e A

which is true. Look at the rewritten statement:

~(x ~e A) ® ~(x e A\A), or

x e A ® x ~e A\A.

Now this holds for any variable. Hence A\A is a set with no elements.

Now suppose B ~= A. Nevertheless,

x e B\B « x e A\A

as we see from

(x ~e B\B or x e A) and (x e B\B or x ~e A),

which is true.

So let us denote A\A with an empty set symbol { } and consider the cases { }\{ }, ~{ }\{ }, and { }\~{ }.


i.    x e { }\{ } --> x e { } and x e { }
          which holds vacuously [note transformations above]

ii.   x e ~{ }\{ } --> x e ~{ } and x ~{ }
          which is true.

iii.  x e { }\~{ } --> x e { } and x ~e ~{ }
          which means
      x e {}\~{} --> x e {} and x e {}
          which is false, meaning that the expression
          {}\~{} is not defined in standard set
theory.

Now we have that A\A = B\B = { }\{ } = { }, establishing that the null subset of any set is indistinct from the null subset of any other set, meaning that there is exactly one null set.


Paul Conant's Erdos number is not a member of Z+.

Email: prconant@yahoo.com