Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

I am trying to prove in a different way then how it was already proved on this website (another question). So yes this is sort of a duplicate.

Claim: $\mathbb{Q}$ does not have the least upper bound property

Let $S = \{r \in \mathbb{Q} : r > 0~~ \text {and} ~~r^2 < 2 \}$. Clearly $S$ is bounded above by 2 and $S$ is nonempty since $1 \in S$.

To prove S has no least upper bound it suffices to prove that if $t$ is an upperbound for $S$ then there exists a upper bound $t' \in \mathbb Q$ such that $t' < t$.

It can then be shown that for $n \in \mathbb{N}$ is sufficiently large then $$\left(t-\frac{1}{n} \right)^2 > 2$$ given that $t^2 > 2$

But I get stuck here. How would I go about proving this is true? any ideas would help. I appreciate it.


share|cite|improve this question
Hint: any strictly increasing continuous function has a continuous inverse function. – Robert Israel Jan 13 '12 at 19:27
Or you can use $\sqrt{t} - \sqrt{s} = \frac{t-s}{\sqrt{t}+\sqrt{s}}$ – Robert Israel Jan 13 '12 at 19:28
@Robert: Your hints seem to require the construction of the reals. Otherwise, for instance, $\sqrt t$ has no meaning. But the failure of the least-upper-bound property is an intrinsic property of the rationals, and can be proved without using real numbers at all. – TonyK Jan 13 '12 at 19:34
up vote 1 down vote accepted

If $t^2 > 2$, then $t^2-2>0$ so there must be a positive integer $n>1$ so that $t^2-2>\frac{1}{n}$.

But $$(t-\frac{1}{2nt})^2 = t^2-\frac{1}{n} + \frac 1{4t^2n^2} > t^2-\frac{1}{n} > 2$$

So $t-\frac{1}{2nt}$ is smaller than $t$ and it's square is still greater than 2.

share|cite|improve this answer
I'm learning how to use the least upper bound propierty. So for example here, if $t$ is the least upper bound then there are two cases, $t$ might be in $S$ or it might not. I don't understand why you don't consider the first case, meaning that might be $t^{2}<2$. Also the case $t^{2}=2$ I guess is because you're assuming as proved that $2$ is irrational. Thank you so much for your help. – Daniela Diaz Jun 28 '13 at 7:06
@DanielaDiaz Yes, we are assuming that $\sqrt{2}$ has been proven to be irrational, or, spoken another way, there is no rational number $t$ such that $t^2=2$. – Thomas Andrews Jun 28 '13 at 11:16
Thanks, Thomas. What about the case $t^{2}<2$. – Daniela Diaz Jun 28 '13 at 11:26

If you substitute 4 for 2 in your proof, then the conclusion will be false: $\{r \in \mathbb{Q} : r > 0$ and $r^2 < 4 \}$ does have a least upper bound in $\mathbb Q$ (because 4 is a perfect square). So somewhere you have to use the fact that if $r \in \mathbb Q$, then $r^2 \ne 2$.

So your proof should have two parts:

  1. Show that if $r \in \mathbb Q$, then $r^2 \ne 2$.
  2. Show that if $r \in \mathbb Q$ and $r^2 > 2$, then $\exists t \in \mathbb Q$ such that $2 < t^2 < r^2$.

If you can prove these (can you?), then the result follows: any upper bound $r$ will satisfy $r^2 > 2$ (because of 1), so it can't be a least upper bound (because of 2).

Edited to reply to comment: OK, if $r^2 > 2$, then $r^2 = 2 + \epsilon$ for some rational $\epsilon > 0$. Now look at $t = r - \epsilon/2r$.

share|cite|improve this answer
Yes, my question got up to part 2 (may have missed some details). I am looking for an idea/headstart on how to go about proving number 2. In fact I am trying to prove this with the inequality (involving n) in my questino. – Tyler Hilton Jan 13 '12 at 19:40
@Tyler: See my edit. – TonyK Jan 13 '12 at 19:57
You really need to look at $r-\frac{\epsilon}{2r}$. – Thomas Andrews Jan 13 '12 at 21:40
@Thomas: Thanks, I fixed it now. – TonyK Jan 14 '12 at 8:44

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.