I am struggling to see where the contradiction lies in my proof. In a previous example, $1/\phi = \phi-1$ where $\phi$ is the golden ratio $\frac{\sqrt{5} + 1}{2}$.

Since I am proving by contradiction, I started out by assuming that $ϕ$ is rational. Then, by definition, there exists $a,b$ such that $\phi = a/b$. After some simple calculations and using the result shown from my previous example, I found that $\phi= b/(a-b)$. I also know that $b < a$ from directly calculating the ratio.

I know there is a contradiction in the result $ϕ = b/(a-b)$ but I cannot see it. Any help would be appreciated.

  • $\begingroup$ You could start by proving that $\sqrt{5}$ is irrational, and then use the fact that sum and product of a nonzero rational and irrational number are irrational. $\endgroup$ Sep 13, 2017 at 16:35
  • 5
    $\begingroup$ Another possible approach: $\phi$ is a root of $x^2 - x - 1$, but the only possible rational roots would be $\pm 1$, for example by using Gauss's lemma on factorization of polynomials in $\mathbb{Z}[x]$ - and those obviously aren't roots. $\endgroup$ Sep 13, 2017 at 19:15

6 Answers 6


Here's one idea that works directly without showing anything about $\sqrt 5$:

We know $\varphi > 1$ so if it is rational, we could write $$ \varphi = \frac{a}{b},$$ where $a > b > 0$ are integers and $\operatorname{gcd}(a,b) = 1$. Then using the relation $\frac{1}{\varphi} = \varphi - 1$ gives $$ \frac{b}{a} = \frac{a - b}{b},$$ which is a contradiction since $\operatorname{gcd}(a,b) = 1$ by construction and $a > b$ (it would be a further reduction of a fraction that we already chose to be completely reduced).

  • $\begingroup$ Excellent solution! $\endgroup$
    – Ken Draco
    Sep 13, 2017 at 17:19
  • $\begingroup$ Because the GCD of the two denominators would have to be greater than 1 if both a and b were rational? $\endgroup$
    – CJ Dennis
    Sep 14, 2017 at 6:18
  • 1
    $\begingroup$ @CJDennis no, but because $\operatorname{gcd}(a,b) = \operatorname{gcd}(b-a,b)$ whence both fractions are in reduced form which is unique, i.e., $a=b$, contradiction. $\endgroup$
    – yo'
    Sep 14, 2017 at 10:13
  • $\begingroup$ Sorry for my ignorance but I am struggling to see why this is a contradiction. We assumed $a\over b$ was fully reduced and then showed ${a-b}\over b$ was fully reduced... how is this a contradiction? $\endgroup$
    – Crystal
    Dec 1, 2019 at 23:13
  • 1
    $\begingroup$ @Thomas: one can think of it like so - by assumption, a > b so let c = a - b. Then we have b / a = c / b. Now, since all are integers, and a > b, we should be able to reduce the fraction on the left side to obtain exactly the fraction on the right be cancelling some terms; that is: b = kc and a = kb. But by assumption gcd(a, b) is 1. $\endgroup$
    – Nitin
    Jan 9, 2020 at 9:17

HINT: $$\sqrt{5}=\frac{2a-b}{b}$$ is a contradiction, since the numbers $a,b$ are rational and $$\frac{2a-b}{b}$$ is also rational and $\sqrt{5}$ is irrational

  • $\begingroup$ This really helps! Thank you very much $\endgroup$ Sep 13, 2017 at 16:41
  • $\begingroup$ ok you are welcome $\endgroup$ Sep 13, 2017 at 16:41

If $\phi=\frac ab$ with $\gcd(a,b)=1$ then $\frac 1{\phi}=\phi-1\implies\frac ba=\frac ab-1\implies b^2=a^2-ab=a(a-b)$

This means $b\mid a(a-b)$ but since $\gcd(a,b)=1$ then $b\mid a-b\implies b\mid a$ which is a contradiction.

  • $\begingroup$ For the uninitiated, what is that vertical line that you use in your conclusion? $\endgroup$
    – Dancrumb
    Sep 14, 2017 at 13:57
  • 2
    $\begingroup$ it means "divides" $\endgroup$
    – zwim
    Sep 14, 2017 at 14:03

Though Nitin showed an excellent solution (much better than mine), I've already typed the routine solution approach. So I'm posting it anyway:

If it is rational, we can write it in the form of the fraction $m/n$ where $m$ and $n$ are integers and have no common factors – always, otherwise it is not rational. So we have $$\frac{\sqrt5+1}{2}=\frac{m}{n}$$ $$\sqrt5=\frac{2m}{n}-1$$ But $\frac{2m}{n}-1$ is rational because $\frac{m}{n}$ is rational. Therefore $\sqrt5$ is rational. But it is not. We have a contradition, so the whole number must be irrational. q.e.d.

If you want to prove that $\sqrt5$ is irrational. Do the same: if it is rational, we can write it in the form of the fraction $m/n$ where $m$ and $n$ are integers and have no common factors (We can factor out any common factors). So we have $$\sqrt5=\frac{m}{n}$$ $$5=\frac{m^2}{n^2}$$ $$5n^2=m^2$$ Therefore $m$ is divisible by $5$ and we can re-rewrite it as $m=5k$. So we have $$5n^2=25k^2$$ $$n^2=5k^2$$ Therefore $n$ is also divisible by 5 and we can re-rewrite it as $n=5p$. Therefore, the fraction $\frac{m}{n}$ can be simplified because the numerator and denominator have a common factor of $5$, which contradicts our assumption that there will be no common factors. This cannot be, so $\sqrt5$ cannot be rational. Therefore it’s irrational. q.e.d.


Suppose $\phi$ is rational. From $\frac{1}{\phi} = \phi - 1$, we see that $\phi$ satisfies the polynomial $\phi^2 - \phi - 1 = 0$. By the rational root theorem , $\phi = p/q$ where $p | -1$ and $q | 1$, forcing $\phi = \pm 1$. But neither of these is a root, contradicting that $\phi$ is rational. Therefore, $\phi$ is not rational.


Another way:

Let´s assume that φ is rational.

a/b=φ is completely reduced (we can do this when it is rational)

b<a (by definition)

a/b = (1+ √5)/2 < ( 1 + √9) / 2 = 2 → a < 2 b → a-b < b

From 1/φ = φ - 1

→ b/a = a/b -1 = (a - b)/b

b/a is completely reduced but is equal to another fraction with both a smaller denominator and a smaller numerator.

This is a contradiction, so φ is not rational.

I explained a bit more in :


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.