13
$\begingroup$

I have good reason to suspect that $\sqrt{2+\frac{1}{n}}$ is irrational for all $n \in \mathbb{Z}^+$ but a proof of this eludes me. I've tried proof by contradiction have had no success. I've also tried induction where clearly the base case is true, i.e. $\sqrt{2+1}=\sqrt{3}$ is irrational, but I haven't been able to show the induction step.

$\endgroup$
  • 3
    $\begingroup$ To prove something is irrational, one must (almost) always use a contradiction proof. $\endgroup$ – Michael Burr Jun 30 '15 at 2:29
  • $\begingroup$ Are you serious ? Elementary scrutiny would have been to try a few perfect squares, starting with $n=4$ ! $\endgroup$ – Yves Daoust Jul 20 '15 at 10:41
25
$\begingroup$

In fact, there are infinitely many $n$ with $\sqrt{2+\frac1n}$ rational.

You need $n$ and $2n+1$ to both be perfect squares, say $y^2$ and $x^2$. Then you are trying to solve:

$$x^2-2y^2=1\tag{1}.$$

Then $n=y^2$.

There are infinitely many solutions to this equation, starting with $(x_1,y_1)=(3,2)$ and $(x_{k+1},y_{k+1})=(3x_k+4y_k,2x_k+3y_k)$. So the first is $n=4$, next $n=144$. Next: $n=4900$.

You can actually get the closed form:

$$n =\left\lfloor\frac{(17+12\sqrt 2)^{k}}{8}\right\rfloor.$$

The equation (1) is called Pell's Equation for $2$.

Indeed, if $D$ is a positive integer, then:

  1. If $D$ is not s perfect square then there are infinitely many positive $n$ such that $\sqrt{D+\frac1 n}$ is rational;
  2. If $D$ is a perfect square, then there is no positive $n$ such that $\sqrt{D+\frac1n}$

This all follows from elementary results about Pell's equation.

$\endgroup$
21
$\begingroup$

Disproof: $n=4 \rightarrow 2+1/4= 2.25 ; \sqrt{2.25} =1.5=3/2$

$\endgroup$
  • 5
    $\begingroup$ Better to just keep fractions - decimal notation often gives the impression of being approximations. $\endgroup$ – Thomas Andrews Jun 30 '15 at 3:07
10
$\begingroup$

Let's start a proof by contradiction to try to show that no solutions exist and see where the proof goes wrong. Suppose that $\sqrt{2+\frac{1}{n}}$ is rational and can be written as $\frac{p}{q}$ where $p$ and $q$ share no factors. Then, we know that $$ 2+\frac{1}{n}=\frac{p^2}{q^2}. $$ By clearing $q^2$ from both sides, we have that $$ 2q^2+\frac{q^2}{n}=p^2. $$ Therefore, $\frac{q^2}{n}$ must be an integer, so $n\mid q^2$.

Now, if $\frac{q^2}{n}\not=1$, then there is some prime $s$ dividing $\frac{q^2}{n}$. Therefore, this prime divides $q$. Then, it follows that this prime divides the LHS so it divides $p^2$, but then $p$ and $q$ share a factor (a contradiction).

Therefore, it must be that $q^2=n$. Therefore, the equality becomes: $$ 2n+1=p^2 $$ (which is the same as $2q^2+1=p^2$).

We now begin the construction: Let us assume that $n$ is a square integer such that $2n+1$ is a square. Let $p=\sqrt{2n+1}$ and $q=\sqrt{n}$. Then $\sqrt{2+\frac{1}{n}}=\frac{p}{q}$ and this is the only way to have a counterexample (as seen in the provided disproof).

$\endgroup$
  • 3
    $\begingroup$ Wow, you do all the work, and I get all the points -- my idea of teamwork ;). $\endgroup$ – Gary. Jun 30 '15 at 2:40
  • $\begingroup$ Cheers! @Gary. But, at the end of the day, one must come up with an example that satisfies the conditions I provided (which you did). $\endgroup$ – Michael Burr Jun 30 '15 at 2:42
  • 3
    $\begingroup$ Isn't it easier to just show that if $p,q$ relatively prime, then $p^2$ and $q^2$ are relatively prime. And $2n+1$ and $n$ are relatively prime, so $p^2=2n+1$ and $q^2=n$. You just need to know that every rational number has a unique representation as a quotient of relatively prime numbers. $\endgroup$ – Thomas Andrews Jun 30 '15 at 2:46
  • $\begingroup$ @ThomasAndrews Nice observation. $\endgroup$ – Michael Burr Jun 30 '15 at 2:50

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.