Prove (or disprove) a correlation between a prime number subset and a maximum $n\in\mathbb N_+$

elementary-number-theoryinequalityprime numbers

EDIT: Further expansion on the topic can be found in this post.


I'm kind of testing my own observation that I came up with but may very well have already been worked on. I basically try to find a correlation between the following two:

Let $P = \{p\in\mathbb N_+ | p~$ is prime$\}$ be a non-empty finite set of different prime numbers. Is it true, that:
$${\displaystyle \sum_{p\in P}\frac{1}{p} > 1 \Longrightarrow \exists n_{max}\in\mathbb N_+:\sum_{p\in P}\lfloor \frac{n}{p}\rfloor} < n$$

(note that the existence of a maximal such $n$ is implied)

I also know for a fact (rather easy to prove) that if ${\displaystyle \sum_{p\in P}\frac{1}{p} \le 1}$, then the inequality on the right hand side holds for infinitely large $n$s, so the implication might as well be an equivalence! But I'm currently struggling with the general proofs.

I was inspired by a certain example with the set $P = \{2, 3, 11, 13\}$, which additionally satisfies the condition:
$${\displaystyle \sum_{p\in P}\frac{1}{p}} = 1 + \prod_{p\in P}\frac{1}{p}$$

so that might as well be a necessary condition for the implication to hold true. I've figured out the solution to this partial case to be $n = 1715$.

And generally I'm really interested in this correlation… Any intel on the topic is greatly appreciated! Thanks in advance!

Best Answer

Let $\sum_P(1/p)=1+\delta$ for some $\delta>0$, and let $m$ be the number of elements of $P$. $$\sum_P[n/p]>\sum_P((n/p)-1)=n(1+\delta)-m=n+(n\delta-m)>n$$ for $n>m/\delta$.

Related Question