How to prove that $\mathrm{Hom}(\mathbb{R})=\mathrm{Aut}(\mathbb{R})$ ?

(We treat it as field homomorphisms. )

I know that $\mathrm{Aut}(\mathbb{R})=\{\mathrm{id}\}$ and $\mathrm{Mon}(\mathbb{R})=\mathrm{Aut}(\mathbb{R})$.

  • $\begingroup$ What is the definitions of $\mathrm{Hom}(\mathbb{R})$ and $\mathrm{Mon}(\mathbb{R})$? $\endgroup$ – Krish Jan 21 '15 at 19:06
  • $\begingroup$ Hint: takes the rationals to themselves, preserves order. $\endgroup$ – André Nicolas Jan 21 '15 at 19:07
  • $\begingroup$ $\mathrm{Hom}(\mathbb{K})$ is a set of all homomorphisms $\mathbb{K}\rightarrow\mathbb{K}$. $\mathrm{Mon}(\mathbb{K})$ is set of all monomorphisms $\mathbb{K}\rightarrow\mathbb{K}$. $\endgroup$ – mikis Jan 21 '15 at 19:10
  • $\begingroup$ @mikis Updated my answer to correct a small flaw in it $\endgroup$ – Olórin Jan 22 '15 at 14:35
  • $\begingroup$ @GeorgesElencwajg I did as you do : erased my comment. ;-) Btw, did you erase your false answer from math.stackexchange.com/questions/1112421/… or do you leave it ? ;-) $\endgroup$ – Olórin Jan 22 '15 at 20:20

Let us show that every morphism of fields $f:\mathbb R\to \mathbb R$ is the identity, namely $f(r)=r$ for all real $r$.

0) Trivially $f(q)=q$ for all $q\in \mathbb Q$

1) Notice that $f$ preserves the order relation in $\mathbb R$ : $x\leq y \implies f(x)\leq f(y)$.
Indeed if $r\geq 0$ we can write $r=\rho^2$ for some $\rho\in \mathbb R$ and then $f(r)=f(\rho^2)=(f(\rho))^2$ .
Since the square of a real number is nonnegative , we see that $f(r)\geq 0$.
But then $x\leq y \implies y-x\geq 0 \implies f(y-x)\geq 0\implies f(y)-f(x)\geq 0\implies f(x)\leq f(y)$
and the order relation in $\mathbb R$ is preserved, as claimed.

2) Given $r\in \mathbb R$ we have for any rational $q,q' \in \mathbb Q$ :
$$q\leq r\implies f(q)=q\leq f(r) \quad\operatorname {and} \quad r\leq q'\implies f(r)\leq f(q')$$ Since $r$ and $f(r)$ have the same position relative to rational numbers they are equal: $f(r)=r$ and we have proved that, as announced, $f$ is the identity.

[By the way the definition of real numbers by Dedekind cuts, a definition not very popular nowadays, is based on the very idea that a real number $r$ is characterized by the rational numbers smaller (resp. larger) than $r$ ]


Let $\varphi : \mathbf{R}\to \mathbf{R}$ be a morphism of fields. The kernel of $\varphi$ is an ideal of the field $\mathbf{R}$, and a (commutative) field $k$ has only two ideal, $(0)$ and $k$, and the kernel of $\varphi$ can't obviously be equal to whole $\mathbf{R}$, as then we would have $\varphi(1)=0$ and at the same time (as $\varphi$ is a morphism of fields) $\varphi (1) = 0$. This shows that the kernel of $\varphi$ is $(0)$, that is, that $\varphi$ is injective.

  • $\begingroup$ Thanks for pointing out the mistake. I've deleted my answer. Nice answer. $\endgroup$ – Krish Jan 22 '15 at 4:51
  • $\begingroup$ @Krish Np ! And thx ! Don't hesitate to upvote if it's nice ;-) More seriously yes, a lot of things (even nastier that this one) can happen in $\textrm{Gal}(\mathbf{R}/\mathbf{Q})$, really. $\endgroup$ – Olórin Jan 22 '15 at 12:59
  • $\begingroup$ @Krish Updated my answer to correct it, as there was a flaw in it. $\endgroup$ – Olórin Jan 22 '15 at 14:26
  • $\begingroup$ @RobertGreen, Thank you very much! $\endgroup$ – mikis Jan 22 '15 at 16:07
  • $\begingroup$ In the first line of the construction, there is a typo: "transcendence basis of $\mathbb R$ over $\mathbb Q$" not "over $\mathbb R$". Also can you give a reference/proof for the claim: "Now, as $\mathbb R$ is algebraic over $K,$ one can (classic fact) extend $F:K \to \mathbb R$ to $\mathbb R.$" I know if $K/k$ is an algebraic extension and there is a map $k \to C,$ where $C$ is an algebraically closed field then it can be extended to a map $K \to C.$ But here $\mathbb R$ is not algebraically closed. May be I'm not thinking correctly or missing something trivial. $\endgroup$ – Krish Jan 22 '15 at 16:59

Hint: Let $\phi: \mathbb R \to \mathbb R$ is a ring homomorphism. Prove the followings: (1). $\phi(r) = r, \forall r \in \mathbb Q.$ (2). $\phi(x) > 0, \forall x >0.$ (3). $|x-y|<\frac{1}{m} \Rightarrow |\phi(x) - \phi(y)|<\frac{1}{m}.$

  • $\begingroup$ Right, it can be proved as $\mathrm{Aut}(\mathbb{R})=\{\mathrm{id}\}$. So, can it be generalize for dense-ordered field $(F,<)$ ? What condition we must have to prove that $\mathrm{Aut}(F)=\mathrm{Mon}(F)$ ? $\endgroup$ – mikis Jan 21 '15 at 19:20
  • $\begingroup$ @mikis: I'm sorry, but I don't know much about ordered field. You can see some information about it in the answers by Pete L. Clark in the following two links: (1) math.stackexchange.com/questions/449404/… (2)math.stackexchange.com/questions/276189/… $\endgroup$ – Krish Jan 21 '15 at 19:52
  • $\begingroup$ @Krish $\mathrm{Hom}(\mathbb{R})=\mathrm{Aut}(\mathbb{R})$ is false, as you may see in my answer. OP never wrote that morphisms were continuous, and even restated - without mention to continuity - what Mon and Hom mean. $\endgroup$ – Olórin Jan 22 '15 at 1:00

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.