MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $\sigma(x) = \sigma_1(x)$ denote the sum of all the positive divisors of $x$.

If $n \in \mathbb{N}$ is odd and $\gcd(n, \sigma(n)) = 1$, then do there exist any solutions to the following equation?

$$2{n^2}\sigma(n) = \sigma({n^2})\sigma(\sigma(n))$$

In other words, does there exist such an odd $N = {n^2}\sigma(n)$ which is also a perfect number?

Here, I also list some (open?) problems mentioned by other researchers:

(1) Suryanarayana: Is it true that every odd perfect number is of the form $m\sigma(m)$ for some odd integer $m$; if so, is $\gcd(m, \sigma(m)) = 1$ necessarily?

(2) M. V. Subbarao: Does every odd perfect number $n$ (if such exist) have the representation $$n = \frac{1}{2}m\sigma(m) \hspace{0.5in} (*)$$

Another question: Whenever $n$ given by $(*)$ is perfect, does it follow that $n$ is odd and $\gcd(m, \sigma(m)) = 1$?

I would appreciate it if anybody could point me to reference(s) to recent results on either Problem (1) or Problem (2).

share|cite|improve this question
recent reference of what? – Franz Lemmermeyer Mar 2 '11 at 19:42
@Franz, by "recent reference" I meant to refer to any new results in the direction of resolving (1) or (2). Editing my question to reflect this. Thanks! – Kashitokiku Teshikiari Mar 2 '11 at 19:51
You would do better by searching and other sources for information on multiperfect and odd perfect numbers, and then asking for a reference request for ONE SPECIFIC PROBLEM if you were worried that you had missed something. If he has the time, I bet Pace Nielsen could more than satisfactorily fill a modest reference request. Gerhard "Baby Don't Let Me Down!" Paseman, 2011.03.02 – Gerhard Paseman Mar 2 '11 at 21:11
In order to have $N=n^2 \sigma(n)$ odd you need $$ n = k^2 $$ (since you want $n$ odd also) for some positive integer $k$, so that you ask for a possible odd perfect number of the form $$ N =k^4 \sigma(k^2) $$ To find any possible odd perfect number seems difficult; one of this more particular form probably is still more difficult to find... – Luis H Gallardo Mar 2 '11 at 21:52
Thank you very much Luis! I didn't realize earlier that parity conditions would settle my original problem. =) Coming up with an answer now. – Kashitokiku Teshikiari Mar 3 '11 at 4:41
up vote 1 down vote accepted

(Thanks to Luis H Gallardo for pointing out the parity condition on $n$.)

(Edited on March 12, 2015)

I was actually trying to (initially) rule out the condition $\sigma(n) = q^k$ for an odd perfect number given in the Eulerian form $N = {q^k}{n^2}$. Just today, I found out that it might be possible to rule out the related condition $\sigma(n) = q$.

That being said, if we plug in $\sigma(n) = q$ into


with the (implicit) constraint $n = m^2$ (and since $n < \sigma(n) = q$ implies Sorli's conjecture that $k = 1$), then we get:

$$1 < I(q) \le \frac{6}{5} < \left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}} < I(n) < I(n^2) < 2$$

where $I(x) = \displaystyle\frac{\sigma(x)}{x}$ is the abundancy index of $x$.

The lower bound $$\left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}} < I(n)$$ was communicated by Ochem to Dris in an e-mail dated April 17, 2013.

(That $\sqrt{I(n^2)} < I(n)$ follows from the inequality $$I(ab) \le I(a)I(b)$$ which is true $\forall a, b \in \mathbb{N}$. Equality holds if and only if $\gcd(a,b)$ = 1.)

Since $n = m^2$, $m$ divides $n$ and $m < n$, so that:

$$\sqrt{I(n)} = \sqrt{I(m^2)} < I(m) < I(m^2) = I(n)$$

But $\left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}} < I(n)$. Thus:

$$\sqrt{\left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}}} < I(m)$$

WolframAlpha gives the approximation $$\sqrt{\left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}}} \approx 1.20183425383797497745284556023594.$$

(Note that we have the equation $I(q^k)I(n^2) = I(q)I(n^2) = I(q)I(m^4) = 2$.)

We therefore have $$I(q) \le \frac{6}{5} < \sqrt{\left(\frac{8}{5}\right)^{\frac{\ln(4/3)}{\ln(13/9)}}} < I(m).$$

Note that $\gcd(q,n)=\gcd(q,m)=1$, so that we have $q \neq m$. I now conjecture that $q < m = \sqrt{n}$ would follow from $I(q) < I(m)$ (and some further related inequalities), so that a contradiction against $n < \sigma(n) = q$ will arise.

share|cite|improve this answer
As you can see, this method can be likened to a form of descent, coupled with exploiting symmetry via the divisibility constraint $\gcd(q, n) = 1$. More on this soon. – Kashitokiku Teshikiari Mar 3 '11 at 5:31

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.