1
$\begingroup$

For $n\in \Bbb N \setminus \{1\}$ we have $n={p_1}^{k_1} \cdots {p_m}^{k_m}$ uniqely, by the fundamental theorem of arithmetic.

We can now define sequence $\alpha(n)$ to be $\alpha(n)=\sum_{r=1}^m k_r$ and take as a convention that $\alpha(1)=0$.

Then we can define sequence $\beta(m)=\sum_{w=1}^m \alpha(w)$.

Does there exist real numbers $q_1$ and $q_2$ such that we have $$\lim_{d \to \infty} \frac {\beta(d)}{{q_1} \cdot{d}^{q_2}}=1$$

$\endgroup$
  • $\begingroup$ I think in the definiton of $\alpha (n)$ you want the upper index of the sum to be $n$? $\endgroup$ – Niklas Sep 26 '17 at 14:47
  • $\begingroup$ @NiklasHebestreit Nope,, all is well-defined here. $\endgroup$ – user480281 Sep 26 '17 at 14:48
0
$\begingroup$

Let $$\Omega(n) = \sum_{p^k \| n} k$$ Then $$ \beta(N)= \sum_{n \le N} \Omega(n) = \Omega(N!)= \sum_{p^k \le N} \lfloor N/p^k \rfloor= N\sum_{p^k \le N} \frac{1}{p^k}+\mathcal{O}(N) = N \log \log N+\mathcal{O}(N) $$ by Mertens 2nd theorem

$\endgroup$
-1
$\begingroup$

$\beta(n)$ is OEIS A022559. An asymptotic expression is given there by Charles R. Greathouse IV: $$\beta(n)=n\log\log n+1.0346n+o(n)$$ So the $q_1$ and $q_2$ as constants in the proposed limit do not exist, since $\beta(n)$ exhibits quasilinear rather than polynomial growth.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy