Everyone with any basic knowledge of number theory knows the classic proof of the irrationality of $\sqrt{2}$. Curious about generalizations using elementary methods, I looked up the irrationality of $\sqrt{3}$, and found the following:

Say $ \sqrt{3} $ is rational. Then $\sqrt{3}$ can be represented as $\frac{a}{b}$, where a and b have no common factors.

So $3 = \frac{a^2}{b^2}$ and $3b^2 = a^2$. Now $a^2$ must be divisible by $3$, but then so must $a $ (fundamental theorem of arithmetic). So we have $3b^2 = (3k)^2$ and $3b^2 = 9k^2$ or even $b^2 = 3k^2 $ and now we have a contradiction.

Such a proof follows the same basic logic as the proof for $\sqrt{2}$, except for using the fundamental theorem of arithmetic to replace and generalize the trivial fact that $n$ is even if $n^2$ is even. However, when I apply this proof format to $\sqrt{4} $ (which is clearly an integer and thus rational) I get the following:

Say $ \sqrt{4} $ is rational. Then $\sqrt{4}$ can be represented as $\frac{a}{b}$, where a and b have no common factors.

So $4 = \frac{a^2}{b^2}$ and $4b^2 = a^2$. Now $a^2$ must be divisible by $4$, but then so must $a $ (fundamental theorem of arithmetic). So we have $4b^2 = (4k)^2$ and $4b^2 = 16k^2$ or even $b^2 = 4k^2 $, which implies that $b=4n$ by the fundamental theorem. Now we have a contradiction (since can note that both $a$ and $b$ are divisible by $4$ and we assumed they were coprime)

This proof is clearly false, yet I fail to see where it differs. Where does it do so?

  • 6
    $\begingroup$ Whereas $2$ and $3$ are primes, $4$ is not a prime. So the claim that $4 \mid a^2$ (i.e. $4$ divides $a^2$) does not imply $4 \mid a$. As a counterexample, $4$ divides $a^2 = 36$, but $4$ does not divide $a = 6$. $\endgroup$ – heropup Apr 18 '16 at 21:04
  • 5
    $\begingroup$ Perhaps more obviously (if we were to take the obvious approach and set $a=2$, $b=1$), $4$ divides $2^2$ but does not divide $2$. $\endgroup$ – Fimpellizieri Apr 18 '16 at 21:07
  • $\begingroup$ The essential property here that $2$ and $3$ possess but $4$ does not is squarefreeness: ihey are composed of distinct prime factors. If $d$ is divisible by any prime squared then it is false that "$d$ divides $a^2$" implies "$d$ divides $a$". $\endgroup$ – Erick Wong Apr 18 '16 at 21:51
  • $\begingroup$ Note regarding above comments: primality and squarefreeness are not sufficient and necessary conditions; not being a perfect square is. The argument that the square root of an integer is irrational applies to any non-perfect square, whether composite or containing a nontrivial square factor. $\endgroup$ – Solomonoff's Secret Apr 18 '16 at 21:57
  • 3
    $\begingroup$ There is a foolproof method of finding at what step a false proof like this fails, which is indicated in Fimpellizieri's comment: you substitute what you know to be a counterexample to the final result, and see at what point you've written down a false statement about your counterexample. You can always do this; it requires essentially no cleverness. $\endgroup$ – Qiaochu Yuan Apr 19 '16 at 19:11

Just because $a^2$ is divisible by $4$, that doesn't mean $a$ is.

  • 2
    $\begingroup$ I'm sorry. I deleted all. Imagine I did not write anything. Sorry again. $\endgroup$ – MathLover Mar 8 '18 at 18:45

Your error is stating that if $a^2$ is divisible by 4 so must $a$ be. The fundimental theorem states if a prime $p $ divides $ab $ then $p $ must divide $a $ or $p$ must divide $b $. That is true because $p $ is indivisable.

But if $p $ is composite it doesn't hold. $p$ could equal $jk $ and $j$ could divide $a $ and $k $ divide $b $. Example: 3 divides 4 times 9 so 3 either divides 4 or 3 divides 9 because 3 is prime. But 6 divides 4 times 9 but 6 neither divides 4 nor 9 but instead 3 divides 9 while 2 divides 4 so 6=2 times 3 divide 4 times 9.

So 4 divides $a^2$ means $2*2$ divides $a*a$ so 2 divides $a $ is all you can conclude with certainty. ... because 4 is not prime.

Actually because 4 is not square free. All of its prime factors must divide into $a$ but the square powers can be distributed among (and are) distributed among the square powers of $a$.


Now If you want to prove that square root of a non perfect square is irrational.

Then you can do this what I wrote below as a proof.

Let $p$ is a non perfect square. We are required to prove $\sqrt{p}$ is irrational. Let $\sqrt{p}$

is rational and of the form $\frac{a}{b}$ where $a\;\;and\;\;b$ are coprime and $b\neq{0}$. Now, we have

$$\sqrt{p}=\frac{a}{b}$$ By squaring both sides we have $$p=\frac{a^2}{b^2}$$ $$\implies{a^2}=p{b^2}$$ We know that $a\;and\;b$ are coprime to each other and $g.c.d(a,b)=1$. So we have $$ax+by=1$$ for some non zero integers $x$ and $y$.Then multiply $\sqrt{p}$ on both sides.Then we get $$\sqrt{p}ax+\sqrt{p}by=\sqrt{p}$$See that $a=\sqrt{p}b$. So we have $$pbx+ay=\sqrt{p}$$ Now see that the L.H.S belongs to integer set whereas the R.H.S belongs to $\mathbb{R}\backslash \mathbb{Z}$.

Hence it is a contradiction.So $\sqrt{p}$ is irrational. May be it will help you.

  • $\begingroup$ It is a nice proof but it really doesn't do anything to answer the OP. The question is not "how cause I prove this square root is irrational?" but "I know this proof is wrong somehow, but where is the error?" $\endgroup$ – Erick Wong Apr 19 '16 at 20:02
  • $\begingroup$ That's because it doesn't mean if 4|a^2,then 4|a.A counter example is a=36.4|36 but 4 doesn't divide 6. $\endgroup$ – sayan kundu Apr 20 '16 at 10:04
  • $\begingroup$ Yes, that was given in the other answers and comments. I just find it odd that you submitted an answer to a different question than the one asked here. $\endgroup$ – Erick Wong Apr 20 '16 at 10:25
  • $\begingroup$ If you look at the proof,at last we get a contradiction.In case of P=4.you will not get contradiction.Hence you will get that sqrt of 4 is irrational which is rational.Thats why i wrote in the last line that it may help you. $\endgroup$ – sayan kundu Apr 20 '16 at 10:32
  • $\begingroup$ If you look at the proof then you will see that you will get that sqrt of 4 is irrational which is not. $\endgroup$ – sayan kundu Apr 20 '16 at 10:47

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.