1
$\begingroup$

Let $s_n$ denote the sum of the first $n$ primes. Prove that for each $n$ there exists an integer whose square lies between $s_n$ and $s_{n+1}$.


I cannot give a proof to this, although I have try on some small examples.

I also notice that $\pi(x)\sim x/\log x$ and there are approximately $\sqrt x$ perfect square smaller than $x$. I have a feel that because $\sqrt x=\mathcal O(x/\log x)$, there will be more perfect squares, two or three, between $s_n$ and $s_{n+1}$ when $n$ gets large.

Any suggestion?

$\endgroup$
  • $\begingroup$ A proof of the proposition is claimed in the comments to OEIS007504 $\endgroup$ – Keith Backman Apr 10 at 18:19
2
$\begingroup$

The sum of the first $n$ primes is

$$ s_n = \frac{n^2}{2}\Big(\log n + \log\log n - \frac32 + \frac{\log\log n}{\log n} - \frac{5}{2\log n} + o(1)\Big) $$

Which gives $$ s_{n+1} - s_n \approx n^2 (\log n + \log\log n - 3/2) > n^2 $$

for all $n > 10$. For $n = 1,2,3$ we can verify it manually. Since the gap between $s_n$ and $s_{n+1}$ jumps by more than $n^2$ the interval $s_n, s_{n+1}$ will contains a square. More specifically we can prove that

$$ s_n < (n+1)^2 < s_{n+1} $$

$\endgroup$
  • $\begingroup$ Is there any more elementary proof? This contains a lot of calculations... Besides, thanks for the information, I will look up the formula! $\endgroup$ – kelvin hong 方 Apr 10 at 9:11
1
$\begingroup$

After some discussion with my professor, we have solved this problem by using some elementary arguments.

We first prove the following lemma.

Lemma : For $n\geq 4$ we have $$s_n<\bigg(\frac{p_{n+1}-1}{2}\bigg)^2.$$ To show this we first see $s_4=2+3+5+7=17$ and $(p_{5}-1)/2=5$, so obviously $17< 5^2$. Now if there is $m\geq 4$ such that $s_m<[(p_{m+1}-1)/2]^2$, then $$s_{m+1}=s_m+p_{m+1}<\bigg(\frac{p_{m+1}-1}2\bigg)^2+p_{m+1}=\bigg(\frac{p_{m+1}+1}2\bigg)^2\leq \bigg(\frac{p_{m+2}-1}{2}\bigg)^2$$ since for primes $\geq 11$, every pair of consecutive primes has least gap equals to $2$, the lemma's proof is completed now.

Now choose $n\geq 4$, let $a$ be the largest integer satisfy $a^2\leq s_n$, then by definition we have $(a+1)^2> s_n$. Now we have $$a^2\leq s_n< \bigg(\frac{p_{n+1}-1}2\bigg)^2\implies a< \frac{p_{n+1}-1}{2}$$ therefore $2a+1< p_{n+1}$. Combining these we have $$(a+1)^2=a^2+2a+1< s_n+p_{n+1}=s_{n+1}$$ So we have proved that $s_n< (a+1)^2< s_{n+1}$.

We have $s_1=2, s_2=5, s_3=10, s_4=17$, so $$s_1< 2^2< s_2< 3^2< s_3< 4^2< s_4$$ completely proved this statement.

$\endgroup$

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.