Around a weak form of the Riemann hypothesis inspired in the relationship between the Stolarsky means and the logarithmic mean

57 Views Asked by At

Robin's equivalence to the Riemann hypothesis can be written as

$$\frac{\sigma(n)-n}{\gamma+\log\log\log n}<M_{\text{lm}}(\sigma(n),n)\tag{1}$$

for enough large $n$ (it is well-know this suitable choice of the integer $n$, see [1]), where $M_{\text{lm}}(a,b)$ denotes the logarithmic mean, $\sigma(n)=\sum_{1\leq d\mid n}d$ is the sum of divisors function and $\gamma$ is the Euler-Mascheroni constant. Using the relationship between the logarithmic mean and the Stolarsky mean one can to propose the problem to calculate the greatest integer $k$ for which

$$\frac{\sigma(n)-n}{\gamma+\log\log\log n}<\left(\frac{k\left(\sqrt[k]{\sigma(n)}-\sqrt[k]{n}\,\right)}{\sigma(n)-n}\right)^{-\frac{k}{k-1}}\tag{2}$$
holds.

The articles from Wikipedia for these means are Logarithmic mean and Stolarsky mean.

Question. I would like to know (approximately) what is the greatest positive integer or real number $k$ for which we can to prove that $(2)$ is true for all integer $n>n_{0}$, being $n_0$ a fixed positive constant (maybe depending of $k$) that is a suitable choice (yours) in your discussion. Many thanks.

Please the answer must be without assumption of additional conjectures, thus unconditionally. I add this requirement since I know what is the relationship between $(2)$ and $(1)$. I hope that my project (I mean the inequality $(2)$) has a good mathematical content.

I can to compare numerically $(1)$ and $(2)$ for segments of positive integers $n$'s and fixed integers $k=4$ or $k=5,6\ldots$ increasing, with my computer.

Remarks (Maybe unrelated to my post). Similar discussions (the path from $(1)$ to $(2)$) can be done from other equivalences to the Riemann hypothesis whenever you can to invok ean argument of positivity: I mean for example that $N_k>\varphi(N_k)$ where $N_k$ denotes the primorial of order $k$ and $\varphi(n)$ is the Euler's totient function or the example $\sigma(n)>\log(H_n)$ where $H_n$ denotes the harmonic number (I evoke Kaneko's claim in last paragraph of [2], I also did calculations with my computer for this last example).

References:

[1] G. Robin, Grandes Valeurs de la fonction somme des diviseurs et hypothèse de Riemann, J. Math. Pures Appl. 63, 187-213, (1984).

[2] Jeffrey C. Lagarias, An Elementary Problem Equivalent to the Riemann Hypothesis, The American Mathematical Monthly, 109, No. 6 (2002), pp. 534-543.

[3] Kenneth B. Stolarsky, Generalizations of the logarithmic mean, Mathematics Magazine. 48 (1975),pp. 87–92.