I'm studying the axioms of Zermelo-Frankel Set Theory at the moment. I already know the following six axioms:

  • The axiom of empty set
  • The axiom of extensionality
  • The axiom of pairing
  • The axiom of infinity
  • The axiom schema of separation
  • The axiom of power set

In the lecture we wanted to construct a model of peano arithmetic on $\omega$ just using the above axioms. We started as follows (the language of Peano Arithmetic is $\mathcal{L}_{PA}=\{0,s,+,\cdot\}$):

$$0^\mathbb{N}:=\emptyset$$ $$s^\mathbb{N}:=\{\langle n,m \rangle\in \omega\times \omega\mid m=n \cup \{n\}\} $$ $$+^\mathbb{N}:=\bigcup\{a\in\mathcal{P}((\omega\times\omega)\times\omega)\mid\varphi(a)\}, $$ where $$\varphi(a)\equiv \forall x\in \omega (\langle \langle x,\emptyset\rangle,x\rangle\in a) \land \forall x\forall y\forall z\forall z'(\langle \langle x,y\rangle,z\rangle\in a\land \langle \langle x,y\rangle,z'\rangle\in a \rightarrow z=z')$$

Why does the following hold? $$\forall x\forall y(x+^\mathbb{N}s^\mathbb{N}(y)=s^\mathbb{N}(x+^\mathbb{N}y))$$

Is the definition of $+^\mathbb{N}$ correct? Because even in "simple" cases like $y=\emptyset$ and $x\in\omega$ I don't know what $x+^\mathbb{N}y$ is.

Edit: I thought about my question again and I agree that there's something wrong with the definition of $+^\mathbb{N}$. Here are my thoughts:

Lets define $$+^\mathbb{N}:=\bigcap\{a\in\mathcal{P}((\omega\times\omega)\times\omega)\mid\varphi(a)\}, $$ where $$\varphi(a)\equiv \forall x\in \omega (\langle \langle x,\emptyset\rangle,x\rangle\in a) \land \forall x\forall y\forall z\forall z'(\langle \langle x,y\rangle,z\rangle\in a\land \langle \langle x,y\rangle,z'\rangle\in a \rightarrow z=z')\land \forall x\forall y\forall z (\langle\langle x,y\rangle,z\rangle \in a \rightarrow \langle\langle x, s^\mathbb{N}(y)\rangle, z\cup \{z\}\rangle \in a)$$

Now I want to prove that $+^\mathbb{N}$ is functional. Therefore let $x,y\in \omega$. If $y=\emptyset$ then $\langle\langle x,y\rangle,x\rangle\in +^\mathbb{N}.$ So assume now that $y\neq \emptyset.$ But then by definition of $\omega$, $y$ is a successor ordinal. So $y=k\cup \{k\}$. If there is a $z_1\in \omega$ such that $\langle\langle x,k\rangle,z_1\rangle\in a$, then $\langle\langle x,y\rangle,z_1\cup \{z_1\}\rangle\in a.$ But if $k\neq \emptyset$, $k$ is again a successor ordinal, etc... We can do that until we get to the empty set. (I'm really not sure about that. I'm still a little bit confused by the definition of the natural numbers.) Uniqueness of the sum of two numbers $x,y\in \omega$ is clear by the condition

$$ \forall x\forall y\forall z\forall z'(\langle \langle x,y\rangle,z\rangle\in a\land \langle \langle x,y\rangle,z'\rangle\in a \rightarrow z=z').$$

  • $\begingroup$ Something is definitely wrong: $\varphi(a)$ just says that $a$ is a binary operation on $\omega$ with $0$ as a right identity, so $+^{\Bbb N}$ as defined here isn’t even a function. Are you sure that you’ve given the full definition of $\varphi$? $\endgroup$ Mar 7, 2015 at 22:59
  • $\begingroup$ I would really like to know why there are two votes to close this perfectly good question. $\endgroup$ Mar 7, 2015 at 22:59
  • $\begingroup$ Thank you for your answer. I gave the definition of $+^\mathbb{N}$ as we wrote it down in the lecture. But I agree that there is something wrong. I thought about my question again. Please see my edited question. $\endgroup$
    – Rungo
    Mar 8, 2015 at 9:12

1 Answer 1


The revised $\varphi$ looks good: $\varphi(x)$ now specifies that $x$ satisfies the minimum requirements to be what we want, and intersecting all such sets ought to cut out all of the extraneous elements. However, your argument that $+^{\Bbb N}$ is functional definitely needs work.

Fix $x\in\omega$. Let

$$B_x=\left\{y\in\omega:\exists u,v\in\omega\left(u\ne v\land\big\langle\langle x,y\rangle,u\big\rangle,\big\langle\langle x,y\rangle,v\big\rangle\in+^{\Bbb N}\right)\right\}\;,$$

and suppose that $B_x\ne\varnothing$. Let $m=\min B_x$, and suppose first that $m=0$. Then there is an $n\in\omega$ such that $\big\langle\langle x,0\rangle,n\big\rangle\in+^{\Bbb N}$ and $n\ne x$. Let

$$a=\left(+^{\Bbb N}\right)\setminus\left\{\big\langle\langle x,0\rangle,n\big\rangle\right\}\;;$$

then $\varphi(a)$, so $+^{\Bbb N}\subseteq a\subsetneqq+^{\Bbb N}$, which is absurd. Thus, $m>0$, and therefore $m=s^{\Bbb N}(k)$ for some $k\in\omega$.

A similar argument works for this case. By the choice of $m$ there is a unique $\ell\in\omega$ such that $\big\langle\langle x,k\rangle,\ell\big\rangle\in+^{\Bbb N}$, but there is an $n\in\omega$ such that $\big\langle\langle x,m\rangle,n\big\rangle\in+^{\Bbb N}$ and $n\ne \ell\cup\{\ell\}$. We can again let

$$a=\left(+^{\Bbb N}\right)\setminus\left\{\big\langle\langle x,m\rangle,n\big\rangle\right\}\;,$$

and we get exactly the same contradiction as in the case $m=0$. It follows that $B_x=\varnothing$ and hence that for each $y\in\omega$ there is a unique $u\in\omega$ such that $\big\langle x,y\rangle,u\big\rangle\in+^{\Bbb N}$. Since $x\in\omega$ was arbitrary, we’ve shown that $+^{\Bbb N}$ is a function.

It remains to check that

$$\forall x\forall y\left(x+^\mathbb{N}s^\mathbb{N}(y)=s^\mathbb{N}\left(x+^\mathbb{N}y\right)\right)\;.$$

The idea is basically the same. Fix $x\in\omega$, and let

$$B_x=\left\{y\in\omega:x+^{\Bbb N}s^{\Bbb N}(y)\ne s^{\Bbb N}\left(x+^{\Bbb N}y\right)\right\}\;.$$

Assume that $B_x\ne\varnothing$, let $m=\min B_x$ and use the definition of $+^{\Bbb N}$ to derive a contradiction. This is very much like what I did above to show that $+^{\Bbb N}$ is a function, so I’ll give you a crack at it first; if you get stuck, leave a comment, and I’ll fill it in.

  • $\begingroup$ Thank you for your answer. It is very helpful. I think that there is a small mistake(?) It should be $a=\left(+^{\Bbb N}\right)\setminus\left\{\big\langle\langle x,m\rangle,n\big\rangle\right\}$ instead of $a=\left(+^{\Bbb N}\right)\setminus\left\{\big\langle\langle x,0\rangle,n\big\rangle\right\}$ (the second time you define a). $\endgroup$
    – Rungo
    Mar 9, 2015 at 18:39
  • $\begingroup$ @Rungo: You’re welcome. You’re right about the mistake: I did a mindless copy/paste of the earlier line and forgot to make the necessary change. Thanks for catching it. $\endgroup$ Mar 9, 2015 at 20:46

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .