Prove that if $p_1,...,p_k$ are distinct prime numbers, then $\sqrt{p_1p_2...p_k}$ is irrational.

I do not usually prove theorems, so any hint is appreciated. I have taken a look at this and tried to repeat that argument over and over, but I messed up. Perhaps there is an easier to do it. Thanks in advance.

Edit: For the case $k=1$, suppose that $\sqrt{p}=\frac{m}{n}, n\neq0$ and $gcd(m,n)=1$. It follows that $m^2=pn^2$, so $p\mid m$. Also, $p\mid n$. Therefore, $m$ and $n$ are not relatively prime. Contradiction.

  • $\begingroup$ Can you write up what you have done? Esp for the $ \sqrt{p_1} $ case. The general case is very similar. $\endgroup$ – Calvin Lin Jun 29 '20 at 22:22
  • 3
    $\begingroup$ it's exactly the same as in the case of two prime numbers that you mention. $\endgroup$ – alphaomega Jun 29 '20 at 22:24
  • $\begingroup$ Just edited to show what I did so far. I'm not so sure that's correct. $\endgroup$ – Mauricio Mendes Jun 29 '20 at 22:38

Assume ${p_1,...,p_k}$ are distinct primes, and assume (aiming for a contradiction) that ${\sqrt{p_1p_2...p_k}=\frac{a}{b}}$ for coprime positive integers ${a,b}$ (alternatively, you can write ${(a,b)=1}$). As before, squaring both sides and rearranging for ${a^2}$ yields

$${\Rightarrow a^2 = p_1...p_kb^2}$$

In other words,${a^2}$ contains ${p_1,...,p_k}$ as factors, and thus ${a}$ must contain ${p_1,...,p_k}$ as factors (since ${a^2}$ contains these primes as factors, this has to be the case because they are prime. This wouldn't be true for some random composite number).

Anyways, we rewrite ${a=p_1...p_ka^*}$. Plugging back in gives

$${\Rightarrow \frac{p_1...p_ka^*}{b}=\sqrt{p_1...p_k}}$$

And this implies

$${\Rightarrow \frac{p_1^2...p_k^2\left(a^*\right)^2}{b^2}=p_1...p_k}$$

You can rearrange this and get

$${b^2 = p_1...p_k\left(a^*\right)^2}$$

And we have got our desired contradiction. By the same argument as before, this would tell us ${b^2}$ has factors ${p_1...p_k}$, and because again these are primes that means ${b}$ contains factors ${p_1...p_k}$. This is a contradiction since we assumed that ${(a,b)=1}$ (that ${a,b}$ were coprime, hence could not share any common factors), and yet from assuming the rationality of our expression we have shown that ${a,b}$ both contained ${p_1,...,p_k}$ as factors!

QED. Or Quantum Electrodynamics if you are a Physicist :P


First, I will prove that if $\sqrt{p_1p_2...p_k}$ is a rational, then it must be an integer. Suppose there are $p, q\in\mathbb{N}$ with $\text{gcd}(p,q)=1$ such that $$\dfrac{p}{q}=\sqrt{p_1p_2...p_k}.$$ Using the Euclid algorithm, we can find two (relatively prime) integers $a, b$ such that $ap+bq=1.$ Now consider $$0=(p-q\sqrt{p_1p_2...p_k})(b-a\sqrt{p_1p_2...p_k})=bp-(ap+bq)\sqrt{p_1p_2...p_k}+aq(p_1p_2...p_k).$$ Hence $\sqrt{p_1p_2...p_k}=bp+aq(p_1p_2...p_k)\in\mathbb{Z}.$

Hence $p^2=p_1p_2...p_k.$ Now from elementary number theory, you can argue that $p_1$ s a prime factor of $p$ and, it implies that $p_1$ is also a prime factor of $p_2p_3...p_k,$ which is a contradiction.

  • $\begingroup$ Note that, this proof does not use any restriction on $k.$ In particular it can be used to derive $\sqrt{p}$ is irrational for any prime $p.$ $\endgroup$ – Bumblebee Jun 29 '20 at 23:27

"and tried to repeat that argument over and over"

that's all you have to do but you only have to do the argument once.

Let $a,b$ be integers where $(\frac ab)^2 = p_1.....p_n$ so

$a^2 = bp_1......p_n$. So for any of those prime $p_i$ (it doesn't matter which one) then $p_i|a^2$ and by Euclid's Lemma[*] the $p_i|a$ so $p_i^2|a^2$ and so $a*\frac {a}{p_i} = b\frac {p_1..... p_n}{p_i}$

For notation let $\frac a{p_i} = a'$ and let $\frac {p_1.....p_n}{p_i} = P= p_1p_2...p_{i-1}p_{i+1}....p_n = \prod_{j=1;j\ne i}^n p_i$.

so $aa' = p_ia'^2 = bP$. So $p_i|bP$.

So by Euclid's Lemma either $p_i|b$ or $p_i|P$. But $P = \prod_{j=1;j\ne i}^n p_i$ is not divisible by $p_i$. So $p_i|b$.

$p_i|a$ and $p_i|b$ so $\frac ab$ where not in lowest terms.

Now we could argue that we only claimed $a,b$ where integer, we never said they had to be in lowest terms. But we can argue if $a,b$ exist then we can repeat this infinitely (as $p_i|b$ then if $\frac b{p_i} = b'$ then we have $(\frac {a'}{b'})^2 = p_1..... p_n$ so we can repeat over and over.) This means we have an infinite series of $ a = p_ia'=p_i^2a_2=p_i^2a_3 =..... =p_k^2a_k=...$. But as $p_i > 1$ this mean $a > a' >a_2>a_3>.....$. That's clearly impossible as there are only a finite number of natural numbers less than $a$[**].

So we have proven the result.

[*] Euclid's lemma: If $p$ is prime and $p|ab$ for integers $a,b$ then either $p|a$ or $p|b$ (or both).

Everything hinges on that. (Which presumably you have proven one way or another already.

[**] This is the well-ordered principal. Every set of natural numbers must have a minimal element. Sa $a> a'> a_2 > a_3.....$ then the set $\{a,a', a_i\}$ must have a smallest $a_k$ which means it only has $k$ elements and eventually we must get $\frac ab$ in lowest terms.


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.