5
$\begingroup$

Let $\ \mathbb N:= \{1\ 2\ \ldots\}\ $ be the set of natural numbers. Let $\ \mathbb P:=\{2\,\ 3\,\ 5\,\ 7\,\ 11\,\ \ldots\}\ $ be the set of primes. Then natural radical $\ rad(n)\ $ is

$$ rad(n)\ :=\ \prod\,\{p\in\mathbb P: p\,|\,n\} $$

for every $\ n\in\mathbb N$. Define

$$ \rho\ :=\ \sum_{N\in\mathbb N}\,\frac 1{n\cdot rad(n)} $$

hence $\ \rho > \frac{\pi^2}6.\ $ Furthermore, it'd be an exercise for 1' year college students (or advanced high school students) to show that $\ \rho<\infty.\ $ Don't tell them too early that

$$\ \rho\ =\ \prod_{p\in \mathbb P}\,\left(1+\frac 1{p\cdot(p-1)}\right) $$

QUESTION $\ $ Can you find (yourself or a reference) an algerbaic expression, possibly in terms of $\pi$ and/or $e$ for $\rho$.

Something like $\ \frac{\pi^{\frac 52}}6$ or similar.


More generally, let

$$ \rho(s)\ :=\ \sum_{n\in\mathbb N}\,\frac 1{n\cdot (rad(n))^s} $$

for every complex $\ s\in\mathbb C\ $ such that it's real part is positive,$\ \Re(s)>0.\ $ Thus

$$ \rho(s)\ =\ \prod_{p\in\mathbb P}\, \left(1+\frac 1{p^s\cdot(p-1)}\right) $$

QUESTION A $\ $What would be the compact expressions for $\ \rho(n)\ $ for all or as many natural values of $\ n\ $ as possible?

QUESTION B $\ $ What are the other formulas for $\ \rho(s)\ $ in the style of multiplicative number theory.

QUESTION C $\ $ How does the complex analytic extension of $\ \rho(s)\ $ look like?


$\qquad\qquad\qquad$ NOTES

Here is my third personal proof (:-) of the infinitude of primes:

THEOREM (Euclid) $\ |\mathbb P| = \infty$

PROOF $\ $ We see from the product representation of $\ \rho\ $ that $\ \rho<\infty\ $ in general, and especially if $|\mathbb P|<\infty.\ $

On the other hand, when $\ |\mathbb P|<\infty$ then $\ r:=\prod\mathbb P\in\mathbb N.\ $ Then

$$ \rho\ >\ \sum_{k=1}^n\,\frac 1{k\cdot rad(k)}\ \ge\ \frac 1r \cdot\sum_{k=1}^n\,\frac 1k\ \longrightarrow\ \infty $$

when $\ n\rightarrow\infty.\ $ Assumption $\ |\mathbb P|<\infty\ $ has lead us to a contradiction.

END of proof

$\qquad\qquad\qquad$ ADDENDUM

Let me copy @Lucia's formula from their first comment below. You may also read Lucia's second comment related to a paper by Bateman. In the derivation below I applied @Wojowu's observation from a comment below, which has drastically simplified and shortened one of the steps.

THEOREM (@Lucia) $$ \rho\ =\ \frac {\zeta(2)\cdot\zeta(3)}{\zeta(6)} $$

PROOF

$$ \frac{p^2}{p^2-1} \cdot \frac{p^3}{p^3-1}\ = \ \frac{p^3+1}{p\cdot(p^2-1)} \cdot \frac{p^6}{p^6-1}\ = $$ $$ \frac{p^2-p+1}{p\cdot(p-1)} \cdot \frac{p^6}{p^6-1}\ = \ \left(1 + \frac 1{p\cdot(p-1)}\right) \cdot \frac{p^6}{p^6-1} $$

i.e. $$ \left(1 + \frac 1{p\cdot(p-1)}\right) \cdot \frac{p^6}{p^6-1}\ = \ \frac{p^2}{p^2-1} \cdot \frac{p^3}{p^3-1}$$

Thus $$ \prod_{p\in\mathbb P}\,\left( \left(1 + \frac 1{p\cdot(p-1)}\right) \cdot \frac{p^6}{p^6-1}\right)\ = \ \prod_{p\in\mathbb P}\, \left(\frac{p^2}{p^2-1} \cdot \frac{p^3}{p^3-1}\right)$$ or $$ \rho\cdot\zeta(6)\ =\ \zeta(2)\cdot\zeta(3) $$

and the theorem follows. END of proof

$\endgroup$
8
  • 15
    $\begingroup$ $\zeta(2)\zeta(3)/\zeta(6)$. $\endgroup$
    – Lucia
    Jan 30, 2016 at 17:21
  • 1
    $\begingroup$ Oooo...o, you @Lucia are so quick! This is very elegant. One cannot even complain about $\zeta(3)$ since it fits the formula so well. I'd like to encourage you to post this comment as an answer (perhaps together with some additional comment within your answer). $\endgroup$ Jan 30, 2016 at 17:38
  • $\begingroup$ @Lucia -- I still hope that you will copy your comment to an "Answer". Otherwise, please turn on the green light for me to insert your formula into the "QUESTION" above (I'd add a derivation too; I'd credit you with your expression, of course). $\endgroup$ Feb 2, 2016 at 18:47
  • 3
    $\begingroup$ I didn't really have anything to add to that one line. By all means include it in the question. (Ok here's one more line to go with my comment: the same constant also appeared in an old asymptotic formula of Bateman to count the number of integers $n$ for which $\phi(n)\le x$. This is why I recognized the Euler product at once, but it is just a coincidence.) $\endgroup$
    – Lucia
    Feb 2, 2016 at 18:50
  • 1
    $\begingroup$ For the proof of Lucia's formula, wouldn't it be enough to note that after using a formula for a sum of geometric series the inequality you state for every $p\in\Bbb P$ is just $(1+\frac{1}{p^2(1-1/p))})\cdot\frac{1}{1-1/p^6}=\frac{1}{1-1/p^2}\cdot\frac{1}{1-1/p^3}$ which is true by straightforward calculation? $\endgroup$
    – Wojowu
    Feb 3, 2016 at 13:30

1 Answer 1

1
$\begingroup$

We already have a definition plus two expressions for a real number $\rho$, including a $\zeta$-formula provided by @Lucia:

$$\rho\ :=\ \sum_{n\in\mathbb N}\,\frac 1{n\cdot rad(n)}\ = \ \prod_{p\in\mathbb P}\,\left(1+\frac 1{p\cdot (p-1)}\right) \ =\ \frac {\zeta(2)\cdot\zeta(3)}{\zeta(6)} $$

Let $\rho$ enjoy another on. In addition to function $rad(n)$ we will need an auxiliary function $rad'(n)$:

$$ rad(n)\ :=\ \prod\{p\in\mathbb P: p|n\}\\ rad'(n)\ := \prod\{p-1: p\in rad(n)\} $$

for all $\ n\in\mathbb N.\ $ Next, let:

$$ \mathbf {Rad}\ :=\ \{n\in\mathbb N: rad(n) = n\} $$

THEOREM

$$ \rho\ =\ \sum_{r\,\in\,\mathbf Rad}\, \frac 1{r\cdot r'} $$

where $\ r' = rad'(r)$.

REMARK 1 $\ rad'(1) = 1\ $ due to the Bourbaki kind of a convention.

PROOF

$$\rho\ :=\ \sum_{n\in\mathbb N}\,\frac 1{n\cdot rad(n)}\ =\ \sum_{r\in\mathbf{Rad}}\,\left(\frac 1{r^2}\cdot\prod_{p|r}\frac p{p-1}\right) $$ $$ =\ \sum_{r\in\mathbf {Rad}}\,\frac 1{r\cdot r'} $$

END of proof

REMARK 2 Hm, equality

$$ \prod_{p\in\mathbb P}\,\left(1+\frac 1{p\cdot (p-1)}\right)\ =\ \sum_{r\,\in\,\mathbf Rad}\, \frac 1{r\cdot r'} $$

is immediate.

$\endgroup$
2
  • $\begingroup$ We see that $\rho$ admits expressions purely via $\mathbf{1.}$natural numbers $n\in\mathbb N,\ $ $\mathbf{2.}$primes $p\in\mathbb P,\ $ $\mathbf{3.}$zeta values of $2\ 3\ 6,\ $ $\mathbf{4.}$radicals $r\in\mathbf {Rad}$. $\endgroup$ Feb 12, 2016 at 0:01
  • $\begingroup$ I'd move this Answer to the Question as an additional remark if it wouldn't overload the Question (I'd have to copy some definitions, etc. from the Answer). $\endgroup$ Feb 12, 2016 at 7:55

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.