After we’ve properly defined the natural numbers, it’s time do define arithmetics on ordinals – such as addition, multiplication and exponentiation. However, we need to do a little bit of work first.
However, what I am going to discuss about today is improtant regardless my goal to define arithmetics.
I am going to show a really strong connection between well-ordered sets and ordinals.
Some reminders
So we have the term of an ordinal which is a -transitive set which is well ordered with respect to the relation
.
We have proved many thing about ordinals, and you can always read everything (with proofs) I’ve discussed about them here.
After we’ve defined ordinals, we’ve met the term of a lower set, which you can read about here.
One theorem that we proved and I want you to remember is: If are both well ordered, then they are isomorphic to each other or one of them isomorphic to a proper lower set of the other.
During the proof of that result, we’ve also proved a really important statement:
If is an order-isomorphism and
, then
.
Ok, enough with the reminders, let’s begin:
Meet the order type
Here is a great theorem:
Supposet that is well ordered set. Them there exists some ordinal
such that: [latexm]A\cong \alpha[/latex].
Why is it so great? Since with this theorem we now know, that any well-ordered set is order-isomorphic to some ordinal. This theorem shows how strong and fundamental ordinals are.
Moreover, Assuming that we proved this theorem, we can conclude even a stronger proposition – This from the theorem is unique. There is only one ordinal a well-ordered set is isomorphic to. Why?
Suppose that is well-ordered and it is isomorphic to both
. Thus:
\alpha\cong A\cong \beta\Rightarrow\alpha\cong\beta
And we’ve proved that if two ordinals are isomorphic to each other, then they are the same – thus, .
The ordinal that is isomorphic to
is called it’s order-type and we denote it by:
\text{otp}(A)=\alpha
That’s a really important term, as we will see in the in the near future.
A small remark
When I first saw this theorem I thought:
“Wait, if any well ordered set is isomorphic to an ordinal – does that mean it is an ordinal itself? I don’t think that makes sense at all, cause I already know well-ordered sets which are not ordinals. Then what is going on here?”
My problem was that I thought that order isomorphism is ‘too strong’. It is indeed preserves the structure of the set, however ordinals are ordered with a specific orde – the order, and it has to be
-transitive. When looking on other well-ordered set, that may not be the case at all!
For example, Let be the set:
A=\{[0,1],[0,2],[0,3]\}
(Those are closed intervals) Where the order is . Of course this set is well ordered, but it is not
-transitive, and not well-ordered with respect to the order
.
On the other hand, it’s not hard to see (and prove) that , which is it’s order type:
\text{otp}(A)=3
Cool, now that this point is clear, we are ready for the proof:
The proof
Let be a well ordered set. We are going to deal pairs of the form:
(x,\alpha)
where and
is an ordinal such that:
x_\downarrow\cong\alpha
Let’s denote the set of all those pairs as . Notice that the set is not empty!
is well ordered, thus it has a first element
. Therefore:
a_\downarrow=\{\emptyset\}=0
Thus .
I am going to prove that we can create an order-isomorphism from the set .
(The proof is going to be really similar to the one where I proved that two well ordered sets are isomorphic to each other or one is a lower set of the other).
Let’s begin with poperties:
is ‘functional’
That is, if then
.
Thats easy, since by the assumption:
x_{\downarrow}\cong \alpha,x_{\downarrow}\cong \beta\Rightarrow \alpha\cong\beta\Rightarrow \alpha=\beta
is ‘one-to-one’
That is, if then
.
By the assumption:
x_{\downarrow}\cong \alpha,y_{\downarrow}\cong \alpha\Rightarrow x_{\downarrow}\cong y_{\downarrow}
Both are well-ordered. Aiming for contradiction, suppose that
. Then, WLOG
(since
is in particular linearly ordered) So:
x_{\downarrow}\subsetneq y_{\downarrow}
So is a proper lower set of
. From that we conclude that they are not isomorphic to each other (a well ordered set is not order isomorphic to a proper lower set of it). And that’s a contradiction to our assumption.
Thus, as we wanted.
is a lower set of 
Suppose that and
. We want to show that
as well.
Since we know that
for some ordinal
.
Let be the order-isomorphism:
f:x_{\downarrow}\to\alpha
thus,
. Using the fact that I mentioned in the beginning we know that:
b_{\downarrow}\cong f(b)_{\downarrow}
However, is an ordinal as an element of one. Thus, it equals to the lower set it genarates:
f(b)_{\downarrow} = f(b)
Therefore, we now have:
b_{\downarrow}=f(b)\Rightarrow(b,f(b))\in F\Rightarrow b\in\text{dom}(F)
is ‘order preserving‘
We need to show that if such that:
(x,\alpha),(y,\beta)\in F, x< y
Then:
\alpha < \beta
But we’ve already seen that in the latest part (try to prove it yourself in a similar way to the previous part – use the isomorphism between and
).
is a transitive set of ordinals
Well, it is already a set of ordinals, we only need to prove that:
\beta\in\alpha\in\text{Im}(F)\Rightarrow\beta\in \text{Im}(F)
Since then there exists some
such that:
(x,\alpha)\in F
That is: . Let
be the isomorphism:
f:x_{\downarrow}\to\alpha
Since and
is onto, there is some
such that:
f(y)=\beta
However, using the fact from the beginning we know that:
y_{\downarrow}\cong f(y)_{\downarrow}=\beta_{\downarrow}\overset{\beta\in\text{ON}}{=}\beta
So .
is an ordinal
A trasitive set of ordinals is an ordinal as well.
We’ll denote it by .
That is, every element of
has some ordinal
such that
.
Suppose that it is not true. Recall that we have proved that is a lower set. Thus, it must be a proper lower set. We’ve proved that any proper lower set in a well-ordered set is generated by an element. Thus:
\text{dom}(F)=a_{\downarrow}\text{ for some }a\in A
In fact, we have proved that is a function from
to
. In addition, it is one-to-one and order preserving from a well-ordered set. Thus, it is an order isomorphism. Therefore:
a_{\downarrow}=\text{dom}(F)\cong\text{Im}(F)
But we’ve also proved that is an ordinal! therefore:
(a,\text{Im}(F))\in F\Rightarrow a\in\text{dom}(F)=a_\downarrow
And that’s obviously a contradiction, since can’t be an element of the lower set genarated by it (a\notin a_{\downarrow}).
Completing the proof
Since is an isomorphism from
to
which is an ordinal, we found that:
A\cong \alpha
As we wanted!
Successor and limit ordinals
After we’ve proved this marvelous theorem, I want to define two types of ordinals.
Recall that a few posts ago (in this one), I’ve defined the ‘function’ , which ‘genarates’ ordinals:
S(\alpha)=\alpha+1=\alpha\cup\{\alpha\}
In fact, an ordinal that we can express as:
s(\beta)=\alpha
for some ordinal is called a successor ordinal.
For example, the natural number are all successors.
On the other hand, which is the set of all the naturals is not a successor as we’ve seen in the post on the natural numbers.
Moreover, 0 is not a successor as well, since it is the first ordinal!
Such ordinals like 0 and – which are not succesors, are called limit ordinals.
(Note that is indeed a successor).
Those terms will play a big role in the future (and in particular, in the next post).
Summary
After meeting the term of an order type of a well ordered set, we are now finally ready to start discussing about arithmetics of ordinals. So that’s my plan for now.
In the next post, I am going to discuss about addition – which is not as simple as you may think…