If it is assumed that $\sqrt{3}$ is known to be irrational (not the case for $\sqrt{5}$), then prove that $\sqrt{3}+\sqrt{5}$ is irrational.

My approach:

Assume that $\sqrt{3}+\sqrt{5}$ is rational. Then there exist coprime integers $p$ and $q$ so that $\frac{p}{q}$ is rational and $\frac{p}{q}=\sqrt{3}+\sqrt{5}$. Thus $(\sqrt{3}+\sqrt{5})^2=2(4+\sqrt{15})=\frac{p^2}{q^2}$, which implies that $p^2$ is even, so $p$ is also even. Let $p:=2m$ for some integer $m$, then $p^2=4m^2$. Thus $q^2(4+\sqrt{15})=2m^2$, which implies that $q^2$ is even, and so $q$ is even. But this contradicts that $p$ and $q$ are coprime, and we arrive at a contradiction.

My way of proving this does not use the fact that $\sqrt{3}$ is irrational. Please let me know if my proof is correct, and how to use the above mentioned fact.

  • 7
    $\begingroup$ $2(4+\sqrt{15})$ is not even an integer, so just because there is a factor of $2$, does not mean that $p^2$ is even. For example, I can write $3 = 2(1.5) = p$, but that doesn't mean $p$ must be even simply because I wrote it as $2 \times 1.5$. $\endgroup$ – heropup Sep 23 '16 at 4:33
  • 1
    $\begingroup$ After you get to $2(4+\sqrt{15})$, you just need to show that $\sqrt{15}$ is irrational. $\endgroup$ – user2825632 Sep 23 '16 at 4:35
  • $\begingroup$ If we assume that $\sqrt{3}+\sqrt{5}$ is rational then its square must also be rational. So we are then assuming that $(4+\sqrt{15})$ is rational. $\endgroup$ – sequence Sep 23 '16 at 4:35
  • 2
    $\begingroup$ This is not the sort of proof where writing a rational number as $p/q$ helps. Just keep track of things like "the sum/product of two rational numbers is rational." $\endgroup$ – Ted Shifrin Sep 23 '16 at 4:36
  • 1
    $\begingroup$ Not completely duplicate, bue a similar method applies: math.stackexchange.com/questions/913934/… $\endgroup$ – Jean-Claude Arbaut Sep 23 '16 at 5:09

This is from here:

Prove that $\sqrt{2}+\sqrt{3}$ is irrational.

More generally, suppose $r =\sqrt{a}+\sqrt{b} $ is rational, where $a$ and $b$ are positive integers.

Then $r(\sqrt{a}-\sqrt{b}) =a-b $ so $\sqrt{a}-\sqrt{b} =\dfrac{a-b}{r} $ is also rational.

Adding and subtracting these, $\sqrt{a}$ and $\sqrt{b}$ are rational.

Therefore, if either or both of $\sqrt{a}$ and $\sqrt{b}$ are irrational, then $\sqrt{a}+\sqrt{b}$ is irrational (and similarly for $\sqrt{a}-\sqrt{b}$).

  • 1
    $\begingroup$ Yes, that's the one method that I used and was thinking about posting :) $\endgroup$ – sequence Sep 23 '16 at 5:43
  • 1
    $\begingroup$ Post first, think later. More points that way:) $\endgroup$ – marty cohen Sep 23 '16 at 6:16
  • $\begingroup$ @martycohen good philosophy. Also let's others think while you thinking. $\endgroup$ – Simply Beautiful Art Sep 24 '16 at 22:10

Let $x=\sqrt 3 + \sqrt 5$. Then:

$$(x - \sqrt 3)^2 = 5$$ $$x^2 - 2 \sqrt 3 x + 3 = 5$$ $$x^2 - 2 = 2 \sqrt 3 x$$ $$(x^2 - 2)^2 = 12 x^2$$ $$x^4 -16 x^2 + 4 = 0$$

By the rational root theorem, if the last equation had rational roots then they would be integer divisors of $4$. But none of $\pm1, \pm2,\pm4$ are roots, so $x$ must be irrational.


If $\sqrt3+\sqrt5=x$ is rational, then $5=(x-\sqrt3)^2=x^2-2\sqrt3x+3$, and


But then $\sqrt3$ is also rational. However, we know that $\sqrt3$ is irrational, so $x$ is also irrational.

And we know $\sqrt3$ is irrational by the usual argument: is $\sqrt3=p/q$, with coprime integers $p$ and $q$, then $p^2=3q^2$. Hence $3$ divides $p$, but then $9$ divides $p^2$ and $3$ divides $q$, so $p$ and $q$ are not coprime, contradiction.


If $\sqrt 3 + \sqrt 5 = p/q; p,q \in \mathbb Z $ then...

$\sqrt 5 = p/q - \sqrt 3$

$5 = (p/q)^2 - 2 (p/q)\sqrt 3 +3$

$2 (p/q)\sqrt 3 = (p/q)^2 - 2$

$\sqrt 3 = (p/2q) - (q/p) $ which is rational. Which is a contradiction as you had already proven that.


However if you hadn't...

This is not the easiest but

If $\sqrt 3 + \sqrt 5 = p/q $

$3 + 2\sqrt {15} + 5 = p^2/q^2$

$\sqrt {15} = p^2/2q^2 - 4 = a/b $

Where $a = p^2- 8q^2; b=2q^2$.

Let $a= 3^j5^k c; 3^l5^m d $ where $c,d $ don't have any factors of $3$ or $5$. ($j,k,l,m $ may be $0$)

$15 =\frac {3^{2j}5^{2k}c}{3^{2l}5^{2m}d}$

$3^{2l+1}5^{2m+1}c = 3^{2j}5^{2k} d$

But the LHS had an odd number of factors of 3 and 5, while the RHS has even (or 0) number of factors of 3 and 5 so they can't be equal.



If $\sqrt{15}=p/q$ is rational and $\sqrt3$ is irrational then $\sqrt5=\frac{\sqrt{15}}{\sqrt3}=\frac{p}{q\sqrt3}$ so $$5=\sqrt5^2=\frac{p^2}{3q^2}.$$ Hence $3$ appears an odd number of times in the prime factorisation of $5,$ a contradiction.

Hope this helps.

P.S. I don't know what I was thinking in putting the old Hint I here.

  • 1
    $\begingroup$ Can you please clarify what you mean by "3 appears an odd number of times in 5"? $\endgroup$ – sequence Sep 23 '16 at 4:53
  • 1
    $\begingroup$ Both $5$ and $\frac{p^2}{3q^2}$ are rational, so we can look at their factorisation into the form $$\prod\limits_ip_i^{n_i}$$ where $p_i$ are primes and $n_i$ might be negative. Since $5=\frac{p^2}{3q^2}$ we conclude that "3 appears an odd number of times in" the factorisation of $5.$ $\endgroup$ – awllower Sep 23 '16 at 5:20
  • $\begingroup$ What's wrong with 3 appearing an odd number of times in 5? $\endgroup$ – sequence Sep 23 '16 at 5:41
  • $\begingroup$ Well, the factorisation of $5$ is unique, and that is $5=5$, with $3$ appearing $0$ times, while $0$ is even. $\endgroup$ – awllower Sep 23 '16 at 6:17
  • $\begingroup$ I) isn't a contradiction. Root (5) is irrational. $\endgroup$ – fleablood Sep 23 '16 at 6:19

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.