0
$\begingroup$

Jeffrey Lagarias proved, unconditionally, that: $$ \sigma(n)<H_n+2\exp(H_n)\log(H_n)\qquad n>1 $$

This was posed as a problem in:

  • J. C. Lagarias, Problem 10949: A generous bound for divisor sums, American Math. Monthly 109 no 6 (2002) 569, doi:10.2307/2695448

(edit: with solution given in AMM 111 no 3 (2004), 264–265, doi:10.2307/4145148)

Where can I find this problem? Or, any other links that shows how the inequality has been derived would be greatly appreciated.

EDIT: I will also accept the answer if anyone can outline the steps, how Lagarias derived his criterion.

$\endgroup$
10
  • 4
    $\begingroup$ It's not a paper. It's a problem Jeff submitted to the Monthly, and you have the citation for the issue and page where the solution was published. What's the problem? $\endgroup$ Commented Oct 13, 2020 at 22:22
  • 1
    $\begingroup$ The Monthly is available through JSTOR jstor.org $\endgroup$
    – Stopple
    Commented Oct 13, 2020 at 22:22
  • 3
    $\begingroup$ You likely were looking at the monthly issue in 2002 (which solely states the problem). The solution can be found in the 2004 issue found here. $\endgroup$ Commented Oct 13, 2020 at 22:29
  • 1
    $\begingroup$ Why is there an 'also' in "Lagarias also proved unconditionally …" in the first sentence of your question? $\endgroup$
    – LSpice
    Commented Oct 13, 2020 at 22:39
  • 1
    $\begingroup$ Please do not vandalize your own questions. $\endgroup$
    – Wojowu
    Commented Nov 14, 2020 at 17:19

1 Answer 1

6
$\begingroup$

The easiest way is to use Ivic's inequality

$\sigma(n)<2.59n \log \log n, n \ge 7$

Then $H_n>\log n +\gamma$ so $e^{H_N}>e^{\gamma}n$ and $\log H_n > \log \log n$ for $n \ge 3$ hence:

$2\exp(H_n)\log(H_n)>2e^{\gamma}n \log \log n>2.59 n \log \log n > \sigma(n), n \ge 7$ and check the cases $n=2,...,6$ by hand

$\endgroup$
2
  • $\begingroup$ Thank you @Conrad $\endgroup$ Commented Oct 13, 2020 at 22:55
  • $\begingroup$ happy to be of help $\endgroup$
    – Conrad
    Commented Oct 13, 2020 at 22:58

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.