I'm a high school student, so please point out my mistakes nicely :)

So we already know odd perfect numbers cannot be in the form of a square, but how about that they cannot be in this form: $$P=abcd...$$ where P, the odd perfect number, equals the product of unique single factors, i.e. a is not b, b is not c, etc. 28, having a prime factorization of $$1 * 2 * 2 * 7$$ is not only not an odd perfect number but has a non-unique factor 2.

So how do we prove that odd perfect numbers cannot have such factors? From the formula for finding the sum of all divisors of a number here, we can deduce the following: $$(a+1)(b+1)(c+1)(d+1)...=2P.$$ Now, are the factors of an odd number even? Of course not. Therefore the factors a, b, c, d, etc, must be odd. Which means we can get the following: $$(E)(Q)(R)(S)...=2P$$ where e, q, r, s,... are even numbers. If there are n factors in P, then: $$\frac{(E)(Q)(R)(S)}{2^n}=\frac{2P}{2^n}$$ therefore $$\frac{(E)(Q)(R)(S)}{2^{n-1}}=\frac{P}{2^{n-1}}$$ Evidently,


Well, unless P only had one divisor, but... then P would equal 1. :P

So please point out any mistakes nicely, please. Thanks! :D

  • $\begingroup$ @peter.petrov Sorry, please look at my edit. $\endgroup$ – HyperLuminal May 28 '15 at 22:13
  • $\begingroup$ The second time you denote by E,Q,R,S different things than the first time. But now it looks more or less OK. I understand it now. The LFH is an integer while the RHS is not unless n=1. Seems correct now, why do you think it's not correct? $\endgroup$ – peter.petrov May 28 '15 at 22:15
  • $\begingroup$ No, I think it's correct :P Is this like not exactly ground breaking or something? $\endgroup$ – HyperLuminal May 28 '15 at 22:19
  • 2
    $\begingroup$ I don't think there is anything wrong with your demonstration. In fact, Euler demonstrated that any odd perfect number $P$ must be of the form $p^{4m+1} Q^2$, where $\lambda$ is a prime of the form $4n+1$. This implies (among other things) that $P$ cannot be square-free, as you yourself have also shown. This is no mean accomplishment, mind you, but it is already known. $\endgroup$ – Brian Tung May 28 '15 at 22:22
  • $\begingroup$ You should continue thinking along these lines and see what else you can come up with! $\endgroup$ – Brian Tung May 28 '15 at 22:23

I think you have the right idea. However, the exposition is insufficiently clear. Let $N$ be an odd perfect number. We show that $N$ is divisible by a perfect square greater than $1$.

Suppose to the contrary that $$N=p_1 p_2\cdots p_n,$$ where the $p_i$ are distinct primes. Then $$2N=(p_1+1)(p_2+1)\cdots (p_{n}+1).$$ This is impossible of $n\gt 1$. For $2^n$ divides the right-hand side, while the highest power of $2$ that divides $2N$ is $2^1$.

We conclude that $n=1$, that is, $N$ is prime. That is impossible, since the sum of the divisors of $N$ would then be $N+1$.


If there are n unique prime factors in P, then (as you noticed)
$(1)\ \ \ E_1 . E_2 . ... . E_n = 2P$,
where $E_i = p_i + 1$, and where $p_i$ is the i-th
(unique) prime factor of P (our perfect odd number).

Now let's denote $M_i = E_i / 2$. The numbers $M_i$ are integers obviously.

From (1) after dividing by $2^n$, we get:
$(2)\ \ \ M_1 . M_2 . ... . M_n = P / (2^{n-1})$

OK, if n > 1, the RHS is not an integer (as P is odd) while the LHS is an integer.
The case n=1 can be considered entirely separately, it's trivial.

  • $\begingroup$ Please see my edit. Can multiple even numbers have a product of a number that is odd*even? $\endgroup$ – HyperLuminal May 28 '15 at 22:14
  • $\begingroup$ That is odd by even - yes. That is odd * 2 -> no of course. $\endgroup$ – peter.petrov May 28 '15 at 22:16
  • $\begingroup$ So what is wrong with my proof? $\endgroup$ – HyperLuminal May 28 '15 at 22:18
  • $\begingroup$ It's just not rigorous enough. But I don't see anything wrong. Read my edits. $\endgroup$ – peter.petrov May 28 '15 at 22:19
  • $\begingroup$ Lol, the other guy who answered @Andre Nicolas did it much better than me. $\endgroup$ – HyperLuminal May 28 '15 at 22:23

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.