11.7 Let $N$ be a number of perfect squares from $\{1,2,...,10^{20}\}$, which 17-th digit from the end is 7, and $M$ be a number of perfect squares from $\{1,2,...,10^{20}\}$, which 17-th digit from the end is 8. Compare $M$ and $N$. (A. Golovanov)
Problem
Source: All-Russian MO Round 4
Tags: inequalities, number theory proposed, number theory
04.04.2005 15:45
Very interesting problem. How many succeded in solving it Myth?
04.04.2005 15:48
How could I know that? BTW, you can send your solution here.
04.04.2005 18:27
Well I didn't solve it to the end but .... The number of perfect squares that have their 17th digit equal to $8$ respectively $7$ are : $M = \sum\([\sqrt{abc9*10^{17}-1}]-[\sqrt{abc8*10^{16}}+1])$ and $N = \sum\([\sqrt{abc8*10^{17}-1}]-[\sqrt{abc7*10^{16}}+1])$ where $abc$ are all numbers with no digit, $1,2,3$ digits. And I am pretty sure that $M > N$
04.04.2005 18:30
Your idea is absolutely correct! But... It seems you did at least 3 mistakes!
04.04.2005 19:46
I think I have finished solving it : We will prove that for each $abc$ a small part from $M$ is greater than its relative from $N$ or I will try to prove that : $\sqrt{10A+10^{17}}+\sqrt{A-10^{16}}-\sqrt{A}-\sqrt{10A} \geq 2$ where $A=abc8*10^{16}$ or if we divide all by $\sqrt{10^{16}}$ we have to prove that : $\sqrt{10B+10}+\sqrt{B-1}-\sqrt{B}-\sqrt{10B} \geq \frac{2}{10^{8}}$ where $B=abc8$. I will prove even more : $\sqrt{10B+10}+\sqrt{B-1}-\sqrt{B}-\sqrt{10B} > 0$ or $\sqrt{10B+10}+\sqrt{B-1} > \sqrt{B}+\sqrt{10B}$ or $9 +2\sqrt{10B^2-10} > 2\sqrt{10B^2}$ or $41 + 40B^2+36\sqrt{10B^2-10} > 40B^2$ which is clearly true so $M > N$
04.04.2005 19:48
Mistakes, mistakes...
09.04.2005 09:38
Where can I get the questions and full solutions to previous Russian MO competitions. The questions are very very interesting. Thank you.
09.04.2005 09:46
You can look for Titu Andreescu book World Math Olympiads (I am not sure in this title).
09.04.2005 10:02
Thank you. I'll look up title. I'm sure I can find it even if information is a little off. Thanks!
09.04.2005 10:26
I found it. Will probably wait until 2004 edition is out since 2003 is the latest so far. Thanks for the recommendation.
09.04.2005 15:44
$\forall n<10^{10}, (n+1)^2-n^2<2.10^{10}+1$ If it exists $a,b,c, N$ such that $abc7\times 10^{16}+N$ is a square, we just have to chose in the $\Big[\frac{abc9\times 10^{16} -abc8\times 10^{16}}{2.10^{10}+1}\Big]$ (at least) squares in $[abc8.10^{16}; abc9.10^{16}-1]$ So informations given in the problem are not enough to have the sign of $M-N$ $280000000^2 < 290000000^2$ $78400000000000000 < 84100000000000000$ $290000000^2 < 420000000^2$ $84100000000000000 < 176400000000000000$
09.04.2005 15:48
09.04.2005 16:19
Oups I had completly misunderstood the problem, now it's ok. I'm sorry
09.04.2005 16:49
Let $N_{abc}$ and $M_{abc}$ the number of perfect squares of the form $abc7\times 10^{16} + Q$ and $abc8\times 10^{16} + R$ with $Q, R<10^{16}$ So $N_{abc}$ is the number of perfect squares in $[abc7\times 10^{16};abc8\times 10^{16}-1]$, in this interval the difference D between 2 consecutives squares verify $2.abc7\times 10^{16}+1\leq D\leq 2.abc8\times 10^{16}-1$ Consequently: $\frac{abc8\times 10^{16}-abc7\times 10^{16}}{2.abc7\times 10^{16}+1}\geq N_{abc}\geq \frac{abc8\times 10^{16}-abc7\times 10^{16}}{2.abc8\times 10^{16}-1}$ $\Leftrightarrow \frac{10^{16}}{2.abc7\times 10^{16}+1}\geq N_{abc}\geq \frac{10^{16}}{2.abc8\times 10^{16}-1}$ using same reasonment for $M_{abc}$ we have: $\frac{10^{16}}{2.abc8\times 10^{16}+1}\geq M_{abc}\geq \frac{10^{16}}{2.abc9\times 10^{16}-1}$ Thus $N_{abc}>M_{abc}$ so $N>M$
09.04.2005 16:53
Inequality $\frac{abc8\times 10^{16}-abc7\times 10^{16}}{2.abc7\times 10^{16}+1}\geq N_{abc}\geq \frac{abc8\times 10^{16}-abc7\times 10^{16}}{2.abc8\times 10^{16}-1}$ is not very good, I think we need to use integer part and add or remove $1$ but the intervall is enough big to keep inequality true
09.04.2005 17:07
I am not sure I understood it
09.04.2005 17:24
Using intuition my reasonment is that between $abc7\times 10^{16}$ and $abc8\times 10^{16} - 1$ there is more perfect squares than between $abc8\times 10^{16}$ and $abc9\times 10^{16} - 1$ since the difference between to consecutives squares is less between $abc7\times 10^{16}$ and $abc8\times 10^{16} - 1$ than between $abc8\times 10^{16}$ and $abc9\times 10^{16} - 1$
11.04.2005 05:19
You know what? I'll let Myth first post if it is correct then I'll rack my brain on it. It looks correct but there is so much too it. Sorry.