14
$\begingroup$

We start with some number $n$ and sum its digits (we can denote sum-of-digits function as $S_d$) to obtain number $S_d(n)$.

If $S_d(n)$ is prime then we calculate number $n^2$ and sum its digits to obtain $S_d(n^2)$. If $S_d(n^2)$ is prime then we calculate number $n^3$ and sum its digits to obtain $S_d(n^3)$, and so on...

We can call number $n$ a power-summer of order m if the numbers $S_d(n),...S_d(n^m)$ are all primes.

We can call number $n$ a power-summer of order $+ \infty$ if $n$ is power-summer of order m for every $m \in \mathbb N$

A question is:

Does there exist a power-summer of order $+ \infty$?

Are you of the opinion that there is some global maximum, that is, a natural number $W$ such that order of every $n$ is less than $W$?

An answer is not in my reach, I do not know much about sum-of-digits functions, but maybe someone has some good ideas.

Peter found a number of order $14$, a number $20619661$ and calculated that upto $n=10^9$ there is no number with an order greater than $14$.

$\endgroup$
  • 1
    $\begingroup$ This might be an interesting question if you edit to include a short table showing the sequence for the first few values of $n$. Do that for several bases. Binary might be the cleanest - then you're just counting the number of $1$s. $\endgroup$ – Ethan Bolker Jun 23 at 14:00
  • 2
    $\begingroup$ According to my program, $$n=20619661$$ has order $14$ $\endgroup$ – Peter Jun 23 at 14:13
  • 1
    $\begingroup$ Yeap, but there is also the case that none of them is true. $\endgroup$ – Βασίλης Μάρκος Jun 23 at 14:56
  • 2
    $\begingroup$ Upto $n=10^9$ , there is no number with an order greater than $14$ $\endgroup$ – Peter Jun 23 at 17:00
  • 1
    $\begingroup$ @Peter I think there is a global maximum. $\endgroup$ – Grešnik Jun 23 at 17:12
1
+50
$\begingroup$

Basic probabilistic heuristics suggest that there is no power-summer of order $+\infty$ nor is there some global maximum $W$.

If we view $n^m$ as a random number between $0$ and $9\log_{10}(n^m) = 9m\log(n)$, then $Pr(S_d(n^m) \text{ is prime}) \approx \frac{1}{S_d(n^m)} \approx \frac{1}{4.5m\log(n)}$. So assuming any slight independence between the events $S_d(n^m)$ (for $n$ fixed as $m$ ranges), which is reasonable, will give that the probability that each $S_d(n^m)$ is prime is $0$.

However, for any given positive integer $W$, $Pr(S_d(n^1),\dots,S_d(n^W) \text{ are prime}) \approx \frac{1}{4.5^W W!}\frac{1}{(\log n)^W}$, so since $\sum_{n \ge 1} \frac{1}{(\log n)^W} = +\infty$, Borel-Cantelli suggests that infinitely many positive integers are power-summer of order $W$.

$\endgroup$
  • $\begingroup$ Can you explain this "Basic probabilistic heuristics suggest that there is no power-summer of order +∞ nor is there some global maximum W."? $\endgroup$ – Grešnik Jun 27 at 8:23
  • $\begingroup$ @Grešnik that's what the rest of my answer does. $\endgroup$ – mathworker21 Jun 27 at 9:11
  • $\begingroup$ But, you assume that "assuming any slight independence between the events $S_d(n^m)$ (for n fixed as m ranges), which is reasonable,", can you explain why do you think that is reasonable? $\endgroup$ – Grešnik Jun 27 at 9:13
  • $\begingroup$ @Grešnik I see no reason why the fact that the sum of the digits of one power of $n$ being prime should influence the fact that the sum of digits of another power of $n$ is prime. Of course, everything here is deterministic, so it is a bit silly to speak in the language of probability and hard for me to explain myself, but that is the best I can do. $\endgroup$ – mathworker21 Jun 27 at 9:32
  • $\begingroup$ @Grešnik But you are right in that I should say "basic probabilistic heuristics" under some sort of independence assumption. I want to emphasize, though, that such an assumption can be very very mild, given how low the probabilities involved are. $\endgroup$ – mathworker21 Jun 27 at 9:34

Your Answer

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

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