Countability
Let me relax the definition of equinumerous a bit. Often I want to compare the sizes of sets without declaring exactly what that size is. I merely want to point out that one set has at least as many elements as another set.
Note:
I give precise intuition for this in the lectures. This requires pointing and sketching so I will not include that here.
Definition.
For two sets and , I use the symbol to indicate that there is an injective function from to .
Here are some basic properties that describe as an order among sets. The same metamathematical warning as above applies here, however.
Proposition.
Let , and be sets.
- If and , then (Transitivity)
- and if and only if (Antisymmetry)
- If , then (Monotonicity)
- (Reflexivity)
- Either or (Linearity)
- if and only if there exists a surjection .
If and , I write .
Stop! Try proving these yourself before continuing or you'll miss out.
It is surprisingly deceptive knowing which proofs are easy and which ones are hard.
Proof:
- Composition of injective functions is injective.
- The implication is known as the Cantor-Schröder-Bernstein theorem, and its proof is beyond the scope of this course. The other implication is immediate.
- The inclusion map is injective.
- The implication of 2 implies this result.
- This is equivalent to the axiom of choice. You will only see this in a set theory course. (But unknowingly use it all the time. E.g. linear algebra, analysis, etc.) Here is a relevant discussion.
- See Problem Set 1.
Definition (Countable).
A set is countable if .
Examples.
The following sets are all countable.
- .
- Any finite set.
- The set of even naturals.
- (Galileo) The set of squares, i.e.
- (Hilbert) .
- .
- The set of positive rationals.
Note:
These proofs are best understood by interactively sketching and drawing, in the lectures.
Proof:
I only do the ones that require work.
- Consider mapping .
- Define as for non-negative and otherwise. Show it is injective.
- Map to . Injectivity follows from FTA. Is this mapping a bijection?
- The same idea as (6) works here too.
In the lecture’s informal discussion, I talked about the concept of cardinal and ordinal numbers. The former are often denoted by the first letter of the Hebrew alphabet, . The smallest transfinite ordinal number is , and the corresponding cardinal is the first infinite cardinal number: . I declare
This section is devoted to convincing you that is the “smallest” infinity.
Theorem.
The set is infinite if and only if .
Proof:
First, suppose that is infinite. I define recursively. Since clearly , pick any element . Suppose that is defined on for all . Since is infinite, . Thus, there exists that is not of the form for any . Let . By induction, is a function . The injectivity of also follows from induction.
Now assume that . Thus there is an injection . This mapping is bijective onto its image, and thus equinumerous with a subset of . If were finite, so would this subset be. But this contradicts that is infinite. Therefore, must be infinite.
Theorem.
A set is countable if and only if it is finite or equinumerous with . In the latter case I say it is countably infinite.
Proof:
This follows from Cantor-Bernstein or “antisymmetry.”
Reflect.
Can you find a set whose powerset is countably infinite?
Theorem ("The cardinal is regular").
The countable union of countable sets is countable.
Proof (Hilbert):
Let be a countable collection of countable sets.
Hint.
If it makes the notation easier for you, assume that and that and rewrite the proof using this notation instead.
By definition, there is an injective mapping , and for every , there is an injective function .
I now define an injective mapping . Recall that is simply the union over all sets in the family, that is
More details on indexed families can be found here. Given , use the well-ordering principle to select the set for which is the smallest natural such that . Then define . Since both and are injective, is injective. Therefore, by the example (6) from the Countability section,
By transitivity, I am done.
Cantor’s theorems
Cantor's diagonal argument.
The set of real numbers is not countable.
Informal proof seen in lectures.
Question (Continuum hypothesis).
Can you find a set such that ?
Answer: In the standard axioms of set theory I follow (Zermelo-Fraenkel or ZFC), this problem is undecidable, meaning both the statement and its negation are proven to be un-provable.
I finally arrive at the first theorem of set theory:
There are infinities of different sizes.
Theorem (Cantor).
For any set , .
An exercise worth writing down before continuing is below.
Exercise.
Formalize and write down an argument that . You may not use this for PS3, as it is the same idea as below, just for functions instead of subsets.
Note:
The diagram seen in lectures should clarify the following argument.
Proof of Cantor's theorem.
First, the mapping is easily seen to be an injection. Thus .
Let be any function. I prove that cannot be surjective and the conclusion follows. The set
is a subset of and hence an element of . Now use the idea behind Russel’s paradox to prove that this set cannot be in the image of .