18
$\begingroup$

I have been asking the following question at MSE with an answer: $\lfloor \sqrt n+\sqrt {n+1}+\sqrt{n+2}+\sqrt{n+3}+\sqrt{n+4}\rfloor=\lfloor\sqrt {25n+49}\rfloor$ is true?

I found this relational expression by using computer. Then, I got interested in the generalization of this expression. After receiving an answer, I decided to start an easy example.

First, I got the following with a proof: $$\lfloor\sqrt n+\sqrt {n+1}\rfloor=\lfloor\sqrt {4n+1}\rfloor=\lfloor\sqrt{4n+2}\rfloor.$$

Note that $\lfloor x \rfloor$ is the largest integer not greater than $x$.

Proof: By the AM-GM inequality, we get $n\lt\sqrt{n(n+1)}\lt n+1/2$. Then, we get $$4n+1\lt\left(\sqrt n+\sqrt{n+1}\right)^2=4n+2-2\left(n+1/2-\sqrt{n(n+1)}\right)\lt4n+2. $$ By the way, there are no perfect squares in $\left(4n+1, 4n+2\right].$ (since $k^2\equiv0, 1(mod 4)$ for any natural number $k$, there is no natural number $k$ such that $k^2=4n+2$.) Now, the proof is completed.

Second, I've just got new relational expressions by using computer: $$\lfloor\sqrt{n}+\sqrt{n+1}+\sqrt{n+2}\rfloor=\lfloor\sqrt{9n+8}\rfloor,$$ $$\lfloor\sqrt{n}+\sqrt{n+1}+\sqrt{n+2}+\sqrt{n+3}\rfloor=\lfloor\sqrt{16n+20}\rfloor$$ for any natural numbre $n$.

I won't ask these expression because I'm going to prove these by myself.

Third, I haven't got any expression about 'six terms expression'. Then, I expect the following theorem would be proven true:

Theorem: For any real number $c$, there exists a natural number $n$ such that $$\lfloor\sqrt{n}+\sqrt{n+1}+\sqrt{n+2}+\sqrt{n+3}+\sqrt{n+4}+\sqrt{n+5}\rfloor\not=\lfloor\sqrt{36n+c}\rfloor.$$

Then, here is my question.

Question: The theorem I wrote is true? If it is true, please show me how to prove it. If it is not true, please give me a counterexample.

Any help would be appriciated.

$\endgroup$
  • 1
    $\begingroup$ Have you been able to adapt the answer that you were given? If not, where did it break down? $\endgroup$ – Calvin Lin Aug 27 '13 at 15:35
18
$\begingroup$

Let $g(n)=\sqrt n+\sqrt{n+1}+\ldots+\sqrt{n+5} $ and $f(n)=\lfloor g(n)\rfloor$. When searching for a $c$ such that $$f(n) = \lfloor \sqrt{36n+c}\rfloor $$ for all $n$, each $n$ gives us some conditions on $c$, namely that $$f(n)\le\sqrt{36n+c}<f(n)+1 $$ that is $$\tag1f(n)^2-36n\le c<(f(n)+1)^2-36n.$$ This way, $n=1$ gives us $64\le c<85$ and $n=11$ gives us $88\le c<133$. Since these inequalites contradict each other, no $c$ works for all $n$, thus your theorem is true.


We can find suitable $c$ if we are allowed to ignore the first few values of $n$:

Note that for $k>0$ $$\sqrt{n+k}-\sqrt{n}=\frac{k}{\sqrt{n+k}+\sqrt n}\in\left(\frac{k}{2\sqrt{n+k}},\frac{k}{2\sqrt{n}}\right),$$ hence $$ 6\sqrt n+\frac{15}{2\sqrt{n+5}}<g(n) <6\sqrt n+\frac{15}{2\sqrt n}$$ and $$\tag236n+90\sqrt{\frac{n}{n+5}}+\frac{225}{4(n+5)}< g(n)^2<36n+90+\frac{225}{4n}.$$ Assume $n\ge20$, so that $4(n+5)\le 5n$. From $\sqrt{1+\frac 5n}<1+\frac{5}{2n}$ we have $\sqrt{\frac{n}{n+5}}>\frac1{1+\frac5{2n}}>1-\frac5{2n}$, so the left hand side of (2) becomes $$ g(n)^2>36n+90\left(1-\frac5{2n}\right)+\frac{45}{n}=36n+90-\frac{180}{n}.$$ Thus for $n>180$ we have $\sqrt{36n+89}<g(n)<\sqrt{36n+91}$ and hence $$ f(n)\in\bigl\{\lfloor\sqrt{36n+89} \rfloor,\lfloor\sqrt{36n+90} \rfloor,\lfloor\sqrt{36n+91} \rfloor\bigr\}.$$ Since $36n+90$ and $36n+91$ cannot be perfect squares ($x^2\equiv 18\pmod{36}$ and $x^2\equiv 19\pmod{36}$ have no solutions), we conclude $$ f(n)=\lfloor\sqrt{36n+89} \rfloor=\lfloor\sqrt{36n+90} \rfloor=\lfloor\sqrt{36n+91} \rfloor$$ at least for $n>180$, and by inspection for all $n>3$.


Remark: The above computations also suggest which value of $c$ we shold check in the general case: Similar as above we find that $$ \sqrt n+\ldots +\sqrt{n+k}\approx (k+1)\sqrt n+\frac{0+1+\ldots +k)}{2\sqrt n}=(k+1)\sqrt n+\frac{k(k+1)}{4\sqrt n}$$ and therefore $$ \left(\sqrt n+\ldots +\sqrt{n+k}\right)^2\approx (k+1)^2n+\frac{k(k+1)^2}{2}.$$ Further details depend on whether the integers near $\frac{k(k+1)^2}{2}$ are squares modulo $(k+1)^2$.

$\endgroup$
  • 1
    $\begingroup$ Thank you very much for a great proof with details. $\endgroup$ – mathlove Aug 28 '13 at 2:59

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.