Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Hi all. My question today will be regarding what I consider to be a "stumbling block" while trying to research odd perfect numbers.

Let $N = {q^k}{n^2}$ be an odd perfect number with Euler prime $q$. Since $\gcd(q, n) = 1$, we know that $q \neq n$.

In 2008, I proved that $q^k < n^2$. This implies that, if $n < q$, then Sorli's conjecture that $k = {\nu}_q(N) = 1$ would follow.

I currently know that $I(q) \leq 6/5 < \sqrt{5/3} < I(n)$, where $I(x) = \sigma(x)/x$ is the abundancy index of $x$. In particular, this means that $$\frac{\sigma(q)}{\sigma(n)} < \frac{q}{n}$$.

Thus, if $q < n$, then $\sigma(q) < \sigma(n)$. (The contrapositive of this last implication is $\sigma(n) < \sigma(q)$ implies that $n < q$.)

Now, since $\sigma(q) = q + 1$, I believe we have three cases to consider:

Case 1: $q < \sigma(q) < n < \sigma(n)$

Case 2: $n < q < \sigma(q) \leq \sigma(n)$

Case 3: $n < \sigma(n) \leq q < \sigma(q)$

I also know that $$\frac{\sigma(q)}{n} \neq \frac{\sigma(n)}{q}.$$

My problem is: How do I dispose of Case 2? The motivation is that I want to establish an equivalence between the inequalities $\frac{\sigma(q)}{n} < \frac{\sigma(n)}{q}$ and $q < n$. This way, all it takes to prove Sorli's conjecture will be to show that $\frac{\sigma(n)}{q} < \frac{\sigma(q)}{n}$.

My idea is to show that $$\frac{\sigma(q)}{\sigma(n)} \leq 1 < \frac{q}{n}$$ cannot occur by considering two separate cases under Case 2:

Case 2A: $\frac{\sigma(q)}{n} < \frac{\sigma(n)}{q}$

Case 2B: $\frac{\sigma(n)}{q} < \frac{\sigma(q)}{n}$

I was wondering if anybody out there would have some (better) ideas on how to improve on the bounds for $\frac{n}{q}$ in Case 2A and for $\frac{\sigma(q)}{\sigma(n)}$ in Case 2B. In particular, a sharp upper bound for $I(n) = \sigma(n)/n$ would be nice!

Edit: (June 9, 2012) If $\frac{\sigma(n)}{q} < \frac{\sigma(q)}{n}$, then $n < q$ and $\sigma(n) < \sigma(q)$, as before. (Edit: Since $I(q) < I(n)$, then $q < n$ implies $\sigma(q) < \sigma(n)$. However, while $\sigma(n) < \sigma(q)$ does imply $n < q$, it can happen that $n < q$ AND $\sigma(q) < \sigma(n)$.) Therefore, we have two cases to consider. Under the first case:

$$\frac{\sigma(n)}{q} < \frac{\sigma(q)}{q} < \frac{\sigma(n)}{n} < \frac{\sigma(q)}{n}.$$

Consequently, $n < \sigma(n) < q < \sigma(q)$, and we have the bounds:

$$\frac{\sigma(n)}{q} < 1$$


$$\sqrt{\frac{5}{3}} < \frac{\sigma(q)}{n}.$$

Under the second case (i.e. $n < q < \sigma(q) \leq \sigma(n)$):

$$1 < \frac{\sigma(q)}{q} \leq \frac{\sigma(n)}{q} < \frac{\sigma(q)}{n} \leq \frac{\sigma(n)}{n} < 2,$$


$$1 < \frac{\sigma(n)}{q} < \sqrt{2}$$ $$\sqrt[4]{\frac{5}{3}} < \frac{\sigma(q)}{n} < 2$$

However, if $\frac{\sigma(q)}{n} < \frac{\sigma(n)}{q}$, then we only know that $\sigma(q) < \sigma(n)$ (from before), and no conclusion can be made about comparing $q$ to $n$, as this scenario falls under two cases:

Case 1: $q < \sigma(q) < n < \sigma(n)$

Case 2: $n < q < \sigma(q) < \sigma(n)$

Under Case 1: $$\frac{\sigma(q)}{n} < 1$$ $$\sqrt{\frac{5}{3}} < \frac{\sigma(n)}{q}$$

Under Case 2: $$1 < \frac{\sigma(q)}{n} < \frac{\sigma(n)}{q} < 2$$ $$1 < \frac{\sigma(q)}{n} < \sqrt{2}$$ $$\sqrt[4]{\frac{5}{3}} < \frac{\sigma(n)}{q} < 2$$

Again, the problem is that the upper bound $I(n) = \frac{\sigma(n)}{n} < 2$ for the abundancy index of the component/divisor $n$ is rather crude, as it only uses the fact that $n$ is a factor of a perfect number and is therefore deficient. I was wondering if anybody out there has some better ideas and/or techniques for improving this particular bound, as such will have a direct bearing on the resulting lower bound for $\frac{n}{q}$ under the case $\frac{\sigma(q)}{n} < \frac{\sigma(n)}{q}$, as in my own previous answer to this question. I hope I have put in sufficient detail for this question. Please do let me know if you need further clarifications/information. Thanks!

share|improve this question
I think the votes to close are premature. Arnie: This would be a better question if you focused in on a particular case. Forget about cases 1 and 3, and maybe even 2b. Give it a little bit of motivation as to where your question comes from, and then ask if anyone knows an improvement of a specific bound. –  Cam McLeman Feb 23 '12 at 15:15
Forgive me if this is an unfair comment, but what I think when I see this question is here is a large amount of high school algebra followed by a request that someone else supply some serious mathematics. –  Gerry Myerson Feb 24 '12 at 11:13
Update: Professor Carl Pomerance pointed me to "a new paper of Acquaah and Konyagin who almost prove that $n < q$ cannot occur". I will post the title of their paper soon. –  Jose Arnaldo Dris Mar 31 '12 at 6:32
The joint paper by Peter Acquaah and Sergei Konyagin is titled On Prime Factors of Odd Perfect Numbers". The abstract says: We prove that a prime factor $q$ of an odd perfect number $x$ satisfies the inequality $q < {(3x)}^{1/3}$." –  Jose Arnaldo Dris Mar 31 '12 at 7:33
I was about to point out a paper on the Odd Perfect Number problem which appeared only yesterday on the ArXiV, but then noticed that Arnie is the author - DOH! In case anyone else is interested, the paper is at arxiv.org/abs/1206.1548. –  John R Ramsden Jun 9 '12 at 9:23
show 13 more comments

1 Answer 1

This is a partial answer to my question, in light of a slight improvement on the lower bound for $\frac{n}{q}$ in Case 2A.

Under Case 2A, we have $\frac{\sigma(q)}{n} < \frac{\sigma(n)}{q}$. Therefore:

$$\frac{q}{n}I(q) < \frac{n}{q}I(n)$$

which implies that:

$${\left(\frac{n}{q}\right)}^2 > \frac{I(q)}{I(n)}.$$

Since $1 < I(q) < I(n) < 2$, we have:

$$\frac{n}{q} > \frac{\sqrt{2}}{2} \approx 0.7071.$$

share|improve this answer
To quote Yemon Choi, who commented on an earlier question you asked (and also answered yourself): This question, and this answer, seem dangerously close to using MO as a kind of public blog or attempted polymath.... –  Woett Mar 11 '12 at 0:20
In case $I(n) < \sqrt{2}$, then the resulting bound here is: $$\frac{n}{q} > \sqrt[4]{\frac{1}{2}} \approx 0.84089641.$$ –  Jose Arnaldo Dris Jun 9 '12 at 16:34
add comment

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.