The isomorphism between two complete ordered fields is unique.

Does my attempt look fine or contain logical flaws/gaps? Any suggestion is greatly appreciated. Thank you for your help!

My attempt:

Let $\mathfrak{R}=\langle \Bbb R,<,+,\cdot,0,1 \rangle,\mathfrak{A}=\langle A,\prec,\oplus,\odot,0',1' \rangle$ be complete ordered fields where $\Bbb R$ is the set of real numbers. Let $\mathfrak{B}=\langle B,\prec,\oplus,\odot,0',1' \rangle$ be the smallest subfield of $\mathfrak{A}$ and $\mathfrak{Q}=\langle \Bbb Q,<,+,\cdot,0,1 \rangle$.

Lemma 1: $\mathfrak{R}$ is isomorphic to $\mathfrak{A}$.

Lemma 2: $\mathfrak{Q}$ is uniquely isomorphic to $\mathfrak{B}$.

By Lemma 1, let $\Phi:\Bbb R \to A,\Psi:\Bbb R \to A$ be isomorphisms between $\mathfrak{R}$ and $\mathfrak{A}$. By Lemma 2, let $f:\Bbb Q \to B$ be the unique isomorphism between $\mathfrak{Q}$ and $\mathfrak{B}$.

Let $X \subseteq \Bbb Q$ be bounded from above and $\sup,\sup'$ supremums w.r.t $<,\prec$ respectively. We next prove that $\Phi(\sup X) = \sup' f[X]$.

$\forall x\in X: x \le \sup X \implies \forall x\in X: \Phi(x) \preccurlyeq \Phi(\sup X) \implies \forall x\in X: f(x) \preccurlyeq \Phi(\sup X) \implies \sup' f[X] \preccurlyeq \Phi(\sup X).$

  • Assume the contrary that $\sup' f[X] \prec \Phi(\sup X)$. Since $B$ is dense in $A$, there exists $b\in B$ such that $\sup' f[X] \prec b \prec \Phi(\sup X)$. Then there exists $p\in \Bbb Q$ such that $f(p)=b$. Thus $\sup' f[X] \prec f(p)=\Phi(p) \prec \Phi(\sup X).$

  • We have $\Phi(p) \prec \Phi(\sup X) \implies p<\sup X \implies p<p'$ for some $p'\in X \implies$ $f(p) \prec f(p')$ for some $p'\in X$ $\implies f(p) \prec \sup' f[X]$. This is a contradiction.

Hence $\Phi(\sup X)=\sup' f[X]$. Similarly, $\Psi(\sup X)=\sup' f[X]$.

Let $X_x=\{p\in\Bbb Q \mid p<x\} \subseteq \Bbb Q$. Since $\Bbb Q$ is dense in $\Bbb R$, $x=\sup X_x$ for all $x\in\Bbb R$. Then $\Phi(x)=\Phi(\sup X_x)=\sup' f[X_x]=\Psi(\sup X_x)=\Psi(x)$ for all $x\in\Bbb R$. It follows that $\Phi=\Psi$.

  • $\begingroup$ You should use that a complete ordered field has a unique ordering, because an element is nonnegative if and only if it has a square root. $\endgroup$
    – egreg
    Jan 17 '19 at 0:05
  • $\begingroup$ Hi @egreg, DanielWainfleet has utilized your idea and posted it as an answer below. I am reading his answer. Have you seen any error in my proof? $\endgroup$
    – Akira
    Jan 17 '19 at 0:10

This is an extended comment on the uniqueness of the isomorphism.

(1a). Let $F, G$ be sub-fields of $\Bbb R$ such that $\psi:F\to G$ is a field-isomorphism. ($\psi$ is not assumed to preserve order.) If $\forall x\in F\,(0\le x\implies \sqrt x\in F) $ then $\psi=id_F$ and $G=F.$

Proof: $\Bbb Q\subset F$ and $\psi|_{\Bbb Q}=id_{\Bbb Q}$ so for any $x\in F$ and any $q\in \Bbb Q$ we have $$x\ge q\iff \psi(x)-q=\psi(x)-\psi(q)=(\psi(\sqrt {x-q}))^2\ge 0\iff$$ $$\iff \psi(x)\ge q$$ so $\Bbb Q \cap (-\infty,x]=\Bbb Q\cap (-\infty,\psi(x)],$ so $x=\psi(x).$

(1b).In particular, letting $F=\Bbb R$ in (1a), the only sub-field of $\Bbb R$ that is field-isomorphic to $\Bbb R$ is $\Bbb R$ itself, and the only field-isomorphism of $\Bbb R$ to $\Bbb R$ is $id_{\Bbb R}.$

(2). If $ B $ is a field and $\psi_1, \psi_2$ are field-isomorphisms from $\Bbb R$ to $B$ then by (1b), $\psi_2^{-1}\psi_1=id_{\Bbb R},$ so $\psi_1=\psi_2.$

  • $\begingroup$ A proper sub-field of $\Bbb R$ can be an ordered field according to an order that is not the usual order $<$ of $\Bbb R$. For example $\{a+b\sqrt 2\,:a,b\in \Bbb Q\}.$ For $a,b\in \Bbb Q$ let $a+b\sqrt 2\,>^*0\iff a-b\sqrt 2<0.$ $\endgroup$ Jan 16 '19 at 23:06
  • $\begingroup$ I think you meant $\psi$ rather than $f$. Please check if my reasoning is correct: For all $x\in F$ and $q\in\Bbb Q$: $\psi(x)-q=\psi(x)-\psi(q)$ [since $\psi|_{\Bbb Q}=\text{id}_{\Bbb Q}$] $=\psi(x-q)$. Then $x \ge q \iff x-q \ge 0 \iff x-q$ $=\sqrt {x-q} \cdot \sqrt {x-q} \iff \psi(x-q)=\psi(\sqrt {x-q} \cdot \sqrt {x-q})=\psi(\sqrt {x-q})\cdot \psi(\sqrt {x-q})=$ $(\psi(\sqrt {x-q}))^2 \ge 0 \iff \psi(x)-q=\psi(x-q) \ge 0$. To sum up, $x\ge q \iff \psi(x)-q \ge 0$ $\iff \psi(x)\ge q$.[...] $\endgroup$
    – Akira
    Jan 17 '19 at 1:54
  • $\begingroup$ [...] It follows that $\{q\in\Bbb Q \mid q\le x\}=\{q\in\Bbb Q \mid q\le \psi(x)\}$. Hence $\sup \{q\in\Bbb Q \mid q\le x\}=$ $\sup \{q\in\Bbb Q \mid q\le \psi(x)\}$ and thus $x=\psi(x)$. As a result, the isomorphism between two fields is unique. $\endgroup$
    – Akira
    Jan 17 '19 at 1:54
  • $\begingroup$ Perfectly correct. And $f$ was a typo for $\psi$. $\endgroup$ Jan 17 '19 at 13:58
  • $\begingroup$ Thank you so much for your verification! Have you seen any error in my proof? $\endgroup$
    – Akira
    Jan 17 '19 at 14:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.