3
$\begingroup$

Consider the sequence $T_n$ formed by combining $n$ and $n^2$ into one number. ie. (A053061) $$T_n=\{11,24,39,416,525,636,749 \cdots\}$$ It is easy to see $$T_n= 10^{\lceil 2 \log_{10}(n) \rceil } n+ n^2$$

I looked at the sequence closely trying to find if there are any perfect squares in the sequence but wasn't able to upto $n=100$. I also was able to prove that if $n^{th}$ term is a perfect square then :

1) $n \equiv 8 \text{ or } 0 (\text{ mod } 9)$

2)In the case where $n=9m$ , m is not a square free number.

But I am unable to attack the question

Do there exist any perfect square in the sequence?

$\endgroup$
10
  • 1
    $\begingroup$ There is something wrong with the formula you gave for $T_{n}$. As it stands, $T_{1} = 2$. On the on the hand, if by $\log n$ you are actually referring to the base $10$ logarithm of $n$ then $T_{n}$ is nothing but $n^{3}+n^{2}.$ $\endgroup$ – user2624 Jun 7 '14 at 16:01
  • 2
    $\begingroup$ Not sure why people are voting to close (unless maybe it was before the post was corrected). The general sequence $B^{\lceil 2\log_B(n)\rceil}\cdot n+n^2$ seems like a perfectly reasonable sequence to study, and asking if it contains squares is also a reasonable question. And I'm pretty sure it's not going to be easy to determine, say, if the sequence contains finitely many or infinitely many squares. $\endgroup$ – Joe Silverman Jun 7 '14 at 16:50
  • 2
    $\begingroup$ $a^2=b^2+10^n b$, so $a^2-(b+10^n/2)^2=10^{2n}/4$. So solutions come from factorizations of $10^n/4$. We are looking for solutions with $ b$ between $10^{n-1}$ and $10^n-1$. This corresponds to factors in a certain range. $\endgroup$ – Will Sawin Jun 7 '14 at 17:44
  • 3
    $\begingroup$ We can remove any extraneous factors of $10$ until $a+b$ and $a-b$ are $2\cdot5^{2n}$ and $2^{2n-3}$. Then clearly neither the sum nor the difference is in the correct range. They are much too big. So I think this is impossible. $\endgroup$ – Will Sawin Jun 7 '14 at 18:11
  • 1
    $\begingroup$ @PerAlexandersson It is not about the age: planetmath.org/florentinsmarandache $\endgroup$ – Benjamin Dickman Jun 8 '14 at 0:07
12
$\begingroup$

There are no squares in this sequence.

Elements of this sequence are of the form $b^2+10^n b$ for $10^{n-1} \leq b^2 < 10^n$. So we are looking for a special solution to the equation $a^2 =b^2+ 10^n b$. This is an easy Diophantine equation to solve: We have $a^2 - (b+10^n/2)^2 = - 10^{2n}/4$, or

$(a+b+10^n/2)(a-b-10^n/2) = -10^{2n}/4$

The two factors on the left side have the same parity. Clearly we may assume $n>1$, so these factors both must be even. If those factors are both multiples of $20$, then $a$ and $b$ are both multiples of $10$, and we may divide $a$ and $b$ by $10$ and subtract $1$ from $n$ without changing the equation. However, it does modify the inequality. So each factor is either a power of $2$ or $2$ times a power of $5$, hence the two factors must be $2 \cdot 5^{2n}$ and $2^{2n-3}$. This gives:

$a+b+10^n/2 = 2\cdot 5^{2n}$

$a-b-10^n/2 = 2^{2n-3}$

$b= 5^{2n} - 2^{2n-4} -10^n/2$

For $n=2$, we get $575$, but $575^2>100$. For higher $n$, it's even worse: $b^2$ is much greater than $10^n$. Similarly, multiplying $a$ and $b$ by $10$ and adding $1$ to $n$ also makes $b^2$ larger, relative to $10^n$. So there are no solutions.

$\endgroup$
3
  • $\begingroup$ can this method be extended tp prove there exist no perfect $k$ powers by combining $n$ and $n^k$ ? $\endgroup$ – Shivam Patel Jun 8 '14 at 4:15
  • $\begingroup$ @ShivamPatel - I don't immediately see how, because the equation is more complicated, but there might be a way. $\endgroup$ – Will Sawin Jun 8 '14 at 12:30
  • $\begingroup$ @ Will Sawin ..the can we proceed to prove that there exist no perfect squares in sequences formed by combining $n$ and $n^k$? $\endgroup$ – Shivam Patel Jun 8 '14 at 12:33

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.