Prove that the only perfect number of the form $2^x 3^y$ is $6$

My proof

A number $n$ is perfect if and only if:

$$\sigma(n)=2n$$ $$\sigma(2^x 3^y)=2^{x+1}3^y$$ $$(2^{x+1}-1)(3^{y+1}-1)=2^{x+2}3^{y}$$

Since $3 \nmid 3^{y+1}-1$ and $2\nmid 2^{x+1}-1$ we'll have that $2^{x+2} \mid 3^{y+1}-1$ and $3^y\mid 2^{x+1}-1$. This means that:

$$\frac{3^{y+1}-1}{2^{x+2}},\frac{2^{x+1}-1}{3^y} \in \Bbb{Z}$$

If $2$ numbers are integer, also their product is integer:

$$\frac{3^{y+1}2^{x+1}-3^{y+1}-2^{x+1}+1}{2^{x+2}3^y} \in \Bbb{Z}$$ $$\frac 32 - \frac{3^{y+1}+2^{x+1}-1}{2^{x+2}3^y} \in \Bbb{Z}$$ $$\frac 12 \left[3-\frac{3^{y+1}+2^{x+1}-1}{2^{x+1}3^y}\right]\in \Bbb{Z}$$

$$\left[3-\frac{3^{y+1}+2^{x+1}-1}{2^{x+1}3^y}\right]\in \Bbb{Z} \ \ (\equiv 0 \pmod{2}) $$

$$z=\frac{3^{y+1}+2^{x+1}-1}{2^{x+1}3^y} \ \ \in \Bbb{Z} \ \ (\equiv 1 \pmod{2})$$

Since the denominator grows more quickly (if $y\neq 0$) this reduces the exercise to a finite number of computations. We can also exclude the cases in which $x=0 \vee y=0$ because:

$$\sigma(2^x)=2^{x+1}-1<2^{x+1}$$ $$\sigma(3^y)=\frac{3^{y+1}-1}{2}<2\times 3^{y}$$

So our computations are reduced to:

$$(x,y)=(1,1) \Rightarrow z=1$$

With the successive couples the denominator becomes bigger than the denominator so it's useless to check. This completes the proof.

Is this correct?

Thank you for your time

  • 1
    $\begingroup$ The Euclid-Euler theorem ought to help tremendously. $\endgroup$ – Arthur May 15 at 15:12
  • $\begingroup$ @Arthur I think that the proof has to be done withouth that superpowerful theorem $\endgroup$ – Eureka May 15 at 15:14
  • $\begingroup$ It is not too difficult to show that the EVEN perfect numbers must be of the form $$2^{n-1}\cdot (2^n-1)$$ where $2^n-1$ is a Mersenne prime. If this is done, the only remaining cases are the powers of $3$ $\endgroup$ – Peter May 15 at 15:18
  • 1
    $\begingroup$ @Eureka It's not superpowerful. See the proof. It's rather elementary. $\endgroup$ – Arthur May 15 at 15:18
  • $\begingroup$ I agree Arthur which does not rule out a short and easy proof with another approach. $\endgroup$ – Peter May 15 at 15:20

If $(2^{x+1}-1)(3^{y+1}-1)=2^{x+2}3^y$, you have equality, not just divisibility.

$3^{y+1}-1=2^{x+2}$ (because those are the only even numbers on both sides of the equation), so $2^{x+1}-1=3^y$.

Following your proof, that means:

$$\frac{(3^{y+1}-1)(2^{x+1}-1)}{2^{x+2}3^y}=\frac12 \left[3-\frac{3^{y+1}+2^{x+1}-1}{2^{x+1}3^y}\right]=1.$$


$$\frac{3^{y+1}+2^{x+1}-1}{2^{x+1}3^y}=1 \Rightarrow 3^{y+1}+2^{x+1}=2^{x+1}3^y+1 \Rightarrow 3^{y+1}=2^{x+1}(3^y-1)+1.$$

I think it's relatively easy to see from this final equation that $x=1$ is the only possible solution, which forces $y=1$. I had a harder time seeing why your fraction $z$ couldn't have other solutions.


Here is a relatively shorter solution to this problem:

Since $N = 2^{p-1}(2^p - 1)$ is an even perfect number whenever $2^p - 1$ is a Mersenne prime, it follows that $\omega(N)=2$ where $N$ is even perfect and $\omega(x)$ is the number of distinct prime factors of $x$.

In our case, $N = 2^x 3^y$. Since $\omega(N)=2$, then $x \geq 1$ and $y \geq 1$.

Suppose that $x > 1$. Then $6 \mid N$ and $6 < N$, which implies that $2 = I(6) < I(N)$, where $I(N) = \sigma(N)/N$ is the abundancy index of $N$ and $\sigma(N)$ is the sum of divisors of $N$. This implies that $N$ is abundant, contradicting the hypothesis that $N$ is perfect.

Similarly, if $y > 1$, then $6 \mid N$ and $6 < N$, which implies that $2 = I(6) < I(N)$. This implies that $N$ is abundant, contradicting the hypothesis that $N$ is perfect.

So it cannot happen that $x > 1$ or $y > 1$. This implies that $x = 1$ and $y = 1$, that is, $N = 6$ is the only perfect number of the form $2^x 3^y$.

  • $\begingroup$ This argument works since every (nontrivial) multiple of a perfect number is abundant. $\endgroup$ – Jose Arnaldo Bebita-Dris Jun 12 at 8:43

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.