Transfinite induction¶
While the class of all ordinals is not a set, it is still transitive and well-ordered by . Regarding the associated order , every set of ordinals has a supremum and (if ) an infimum , which is the smallest element of . Such a smallest element exists actually for every (non-empty) class (since if , we only need to find the infimum of the set of ordinals .) This allows us to prove properties about all ordinals by induction.
We have repeatedly used induction already for ordinals , the first uncountable ordinal.
To prove this principle simply observe that if failed there would have to be a smallest with , contradicting the induction hypothesis.
Since every ordinal is either 0, a successor, or a limit ordinal, we have the following variant of induction.
(i) and (ii) coincide with the usual induction scheme for natural numbers. To cover all ordinals we need to add (iii).
Ordinal recursion¶
The induction principle can be used to define functions by recursion. For example, addition on the natural numbers is given by
In the case of ordinals, we have to consider the limit case, too.
Proof
The uniqueness of the function follows by induction.
To show the existence of , we define the following:
- Call tame if
- Say is compatible with if
It follows by induction that any two tame functions are compatible.
This lets us define the desired as
Then is a function (otherwise there would be two incompatible tame functions), its domain is transitive, and satisfies the recursion condition (since it is the union of tame functions).
It remains to show that is defined on all of . If , then we would have for some ordinal . In particular is a set therefore is a set, for some tame . This could be extended to a tame , contradiction.
Note that we defined explicitly as a union of all partial solutions to the recursion equation.
As with induction, we have the following variant of the recursion principle.
We can establish a similar principle for a well-ordering on a class . In case of a proper class, though, we have to require that for every , the class of all predecessors of
is a set (if is a set, this follows automatically by Separation). If this is the case, the recursion principle yields a function such that
Recursion for well-founded relations¶
More generally, we can define induction and recursion on well-founded relations. We already encountered those in a previous chapter.
If is a set, the minimality condition is again automatically satisfied by Separation.
The set condition allows for taking the -transitive closure of a set : the smallest superset of that is -transitive:
This is done by recursion over the natural numbers. The following is an important example.
We can use the existence of as a set to strengthen the minimality condition to subclasses, similar to the case of the well-ordering of :
To prove this lemma, simply pick any , take its transitive -closure, and intersect it with :
is a set, and by the minimality condition has an -minimal element . has to be minimal for , too, since otherwise there exists with . Since , , and therefore , contradicting the minimality of .
The lemma implies a corresponding induction principle for well-founded relations:
This in turn yields the following.
The Von Neumann hierarchy¶
Is there a way to systematically build the , the universe of all sets, “from below”?
We start with the empty set:
Given , the Power Set axiom requires the set of all subsets to exist, so we set
Finally, at limit stages we simply collect all sets we have obtained so far:
What we really are doing here is to construct a function by ordinal recursion. Think .
The following facts about the hierarchy are easily verified by transfinite induction:
Remarkably, if we assume the axiom of Foundation, we reach all sets this way.
Proof
Let be the class of all sets not in any . Since is well-founded, if is non-empty, it has a -minimal element . This implies that for all , . Define a function by mapping each to the least so that . Since is a set, is a set of ordinals, by Replacement. This set of ordinals has a supremum, say . Then and therefore,
Hence must be empty, and the theorem follows.
We can split the question of “how large” is into two sub-questions:
- How “long” is , that is, how many ordinals are there? Axioms for large cardinals attempt to extend this “length” as far as possible.
- How “wide” is , that is, how large is the power set of a set? A rather “slim” universe is given by the constructible sets, which we will encounter soon.