MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $(A, +, \preceq)$ be an ordered group, namely $(A, +)$ is a group and $\preceq$ is a total order on $A$ such that $x + z \prec y + z$ and $z + x \prec z + y$ for all $x,y,z \in A$ with $x \prec y$.

Q1. If $x,y,z \in A$ and $x \preceq z$, is it true that $x+y \ne y + z$ unless $x = z$?

As pointed out by Yves Cornulier in the comments below, the answer is yes iff the ambient group is abelian. This means that I was too hasty (and optimistic) with Q1, since my original question would have been:

Q2. If $x,y \in A$ and $n$ is an integer $\ge 2$, is it true that $x+y \ne y + nx$ unless $x = 0$?

A couple of observations: The answer is yes if (i) $x+y\preceq y+x$, or (ii) the subgroup, $S$, generated by $x$ and $y$ is Archimedean (this follows from a 1902 theorem by O. Hölder).

Here is an informal reasoning in support of a positive answer: Assume to the contrary that $x+y=y+nx$ for some $n \in \mathbb{N}^+$. Then, $-ky+x+ky = n^kx$ for all $k \in \mathbb{N}$. This shoud be in contradiction to the fact that for $n \ge 2$ and $x \ne 0$ the "gap" between $n^kx$ and $-ky+x+ky$ "grows" larger and larger as $k \to \infty$, while the "distance" between $x$ and $-ky+x+ky$ is "uniformly bounded". The point is that I don't see how to turn the above into rigorous arguments, and particularly how to give a formal meaning to words or expressions like "gap", "grows", "distance", and "uniformly bounded". I had thought of introducing a group norm on $S$, but it didn't help much.

Motivation. I was led to Q2 while working at the proof of some estimates on sumsets in ordered semigroups / monoids.

share|cite|improve this question
Are you writing the group additively, even though it is not abelian? – Tobias Kildetoft Oct 28 '13 at 10:45
Yes, I am. This is common in additive theory (e.g., see Ruzsa's survey Sumsets and structure). – Salvo Tringali Oct 28 '13 at 10:52
This is true only if $A$ is abelian. Otherwise, there exist two distinct conjugate elements $x,z$; since your order is total, we can suppose $x\preceq z$. So there exists $y$ such that $xy=yz$ (I avoid your additive notation which is confusing to me). – YCor Oct 28 '13 at 11:58
Right, I was too hasty (and optimistic). But then, what if $z = 2x$? Actually, this is the case I am interested in (I edited the OP according to your comment). – Salvo Tringali Oct 28 '13 at 13:12
up vote 4 down vote accepted

As I pointed out in the comments, Q1 has a negative answer in general and actually has a positive answer only for abelian groups. Your second question (in the version of October 28 2013, 14:54 GMT) also has a negative answer. Indeed, the Baumslag-Solitar group $G=\langle x,t\mid txt^{-1}=x^n\rangle=\mathbf{Z}[1/n]\rtimes_n\mathbf{Z}$ is bi-orderable if $n\ge 1$ (i.e. has a left and right invariant total order, what is sometimes called orderable, the latter sometimes also meaning left-orderable). Indeed, if you consider the set $P$ of $(y,m)\in\mathbf{Z}[1/n]\rtimes\mathbf{Z}$ such that either $m>0$ or $m=0$ and $y\ge 0$, then $P$ is invariant under conjugation (because $n\ge 1$), is stable by multiplication, and $P\cup P^{-1}=G$, $P\cap P^{-1}=\{1\}$. Hence $P$ defines a bi-invariant order (with $P\smallsetminus\{1\}$ as set of positive elements. But the element $x=(1,0)$ is conjugate to $x^n$, so this is a counterexample to your expectation as soon as $n\ge 2$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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