0
$\begingroup$

(Note: This question was cross-posted from MSE per Dris's request.)

Let $N = q^k n^2$ be an odd perfect number in Eulerian form. (That is, $q$ is prime with $q \equiv k \equiv 1 \pmod 4$ and $\gcd(q,n)=1$. Note that $q \geq 5$.) The Descartes-Frenicle-Sorli conjecture for odd perfect numbers predicts that $k=1$ always holds.

I believe it is currently unknown whether $k$ can be bounded from above, but the following appears to be a viable approach for this problem:

Consider the expression $$N - (q^k + n^2) + 1 = (q^k - 1)(n^2 - 1) = (q - 1)\sigma(q^{k-1})(n + 1)(n - 1).$$

The reason for the $\sigma(q^{k-1})$ term is because of the following expression in this preprint, which gives: $$\frac{\sigma(n^2)}{q^k}=\frac{2n^2}{\sigma(q^k)}=\gcd(n^2,\sigma(n^2))=\frac{2n^2 - \sigma(n^2)}{\sigma(q^{k-1})}$$ which would then be equal to $$\frac{(q - 1)(n + 1)(n - 1)(2n^2 - \sigma(n^2))}{N - (q^k + n^2) + 1},$$ from which it follows that $$\frac{\sigma(n^2)}{q^k}\cdot\bigg[N - (q^k + n^2) + 1\bigg]=\frac{2n^2}{\sigma(q^k)}\cdot\bigg[N - (q^k + n^2) + 1\bigg]$$ $$=\gcd(n^2,\sigma(n^2))\cdot\bigg[N - (q^k + n^2) + 1\bigg]=(q-1)(n+1)(n-1)(2n^2 - \sigma(n^2)).$$

Now, it is conjectured that $q^k < n$ (see this M. Sc. thesis and this paper). This will then imply that $$\frac{2n^2}{\sigma(q^k)}>\frac{8n^2}{5q^k}>\frac{8n}{5}>\frac{8}{5}\cdot{\sqrt[3]{N}}$$ since $\sigma(q^k)/q^k < 5/4$ holds for all $k$. Hence, by using Ochem and Rao's lower bound for the magnitude of an odd perfect number, we get that $N>{10}^{1500}$, so that $$\frac{2n^2}{\sigma(q^k)}>\frac{8}{5}\cdot{{10}^{500}}$$

Notice that in the equation $$\frac{2n^2}{\sigma(q^k)}\cdot\bigg[N - (q^k + n^2) + 1\bigg]=(q-1)(n+1)(n-1)(2n^2 - \sigma(n^2))$$ the RHS does not involve $k$. Holding $q$ and $n$ constant (i.e. considering a specific Euler prime $q$ and a specific square root of non-Euler part $n$), and allowing $k$ to vary, then we have $$\frac{2n^2}{\sigma(q^k)}\cdot\bigg[N - (q^k + n^2) + 1\bigg]=(q-1)(n+1)(n-1)(2n^2 - \sigma(n^2))$$ $$> \frac{8}{5}\cdot{\sqrt[3]{N}}\cdot\bigg[N - (q^k + n^2) + 1\bigg]$$ $$\lim_{k \to \infty}{\bigg(\frac{8}{5}\cdot{\sqrt[3]{N}}\cdot\bigg[N - (q^k + n^2) + 1\bigg]\bigg)} \leq (q-1)(n+1)(n-1)(2n^2 - \sigma(n^2)),$$ from which we get the "contradiction" $$(q-1)(n+1)(n-1)(2n^2 - \sigma(n^2)) \geq \lim_{k \to \infty}{\bigg(\frac{8}{5}\cdot{\sqrt[3]{N}}\cdot\bigg[N - (q^k + n^2) + 1\bigg]\bigg)} \to \infty.$$

Here are my questions:

(1) Is there indeed a contradiction in the last step of this "proof" that $k$ is bounded from above?

(2) Is it possible to make the argument in this "proof" more rigorous?

$\endgroup$
2
  • $\begingroup$ Why was this downvoted? I asked Antalan to cross-post this question from MSE for me because I currently could no longer post questions in MO. $\endgroup$ May 28, 2017 at 5:50
  • $\begingroup$ Additionally, kindly take note of my edit to the question. $\endgroup$ May 28, 2017 at 5:51

1 Answer 1

1
$\begingroup$

Let $N = q^k n^2$ be an odd perfect number with Euler prime $q$.

Unconditionally, it is known that $q^k < n^2$ [Dris, 2012]. This implies that $k$ and $n$ are dependent, which means that the proof in the question is flawed.

(This answer is based on JonMark Perry's in the hyperlinked MSE question.)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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