3
$\begingroup$

Prove that there are infinitely many systems of $23$ consecutive integers whose sum of squares is a perfect square.

My try: $$(n-11)^2+\cdots+(n+11)^2=23n^2+1012=23(n^2+44)=m^2$$ so $m=23k$ , $n^2=23k^2-44$. From $\mod 23$, I see $n=23l+5$ or $n=23l+18$ but i don't know what now.

| cite | improve this question | |
$\endgroup$
  • $\begingroup$ Are you familiar with Pell equations? $\endgroup$ – Erick Wong Jan 5 '15 at 0:37
  • $\begingroup$ It's of "Pell type" but the right side isn't $1$. That is, it isn't as routine to solve $m^2-23n^2=1012$ as it is to solve the same with the right side $1$ $\endgroup$ – coffeemath Jan 5 '15 at 0:40
  • $\begingroup$ I'm not familiar with Pell equations. I look for elementary proof. $\endgroup$ – Sinister Jan 5 '15 at 0:48
1
$\begingroup$

Dario Alpern's solver reports $n=18, k=4$ and $n=28, k=6$ as solutions, then if $(x,y)$ is a solution, so is $(24x+115y,5x+24y)$. It will show you the steps if you ask.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ There is also seed solution $n=28, k=6;$ there are two orbits giving solutions, under the automorphism group of the form. $\endgroup$ – Will Jagy Jan 5 '15 at 0:55
  • $\begingroup$ @WillJagy: The solver reported $n=18,k=-4$. I didn't notice that it generated a new list. $\endgroup$ – Ross Millikan Jan 5 '15 at 1:03
  • $\begingroup$ Anyway, by Cayley-Hamilton, we get two sequences of $n$ values under $n_{j+2} = 48 n_{j+1} - n_j,$ one is $18, 892, 42798...$ the other is $28,1362,65348...$ $\endgroup$ – Will Jagy Jan 5 '15 at 1:04
  • $\begingroup$ Ross, yes, unless the target number is $\pm 1$ we expect to get a separate list from that, and here the target is $-44.$ If the target has more prime factors that are represented by some (primitive) form of the same discriminant, we expect even more lists if there are any. $\endgroup$ – Will Jagy Jan 5 '15 at 1:07
  • $\begingroup$ Checked some things, several orbits for $x^2 - 23 y^2 = 154$ $\endgroup$ – Will Jagy Jan 5 '15 at 1:15
1
$\begingroup$

EDIT, March 2016. Based on what people seemed to want in a recent question on Pell's equation, I wrote a program that solves $x^2 - d y^2 = k$ quite quickly, and identifies the "fundamental" solutions, from which all others can be found by applying the automorphism group.

jagy@phobeusjunior:~$ ./Pell_Target_Fundamental

  24^2 - 23 5^2 = 1

 x^2 - 23 y^2 = 154

Thu Mar 31 10:59:54 PDT 2016

x:  19  y:  3 ratio: 0.157895  fundamental 
x:  27  y:  5 ratio: 0.185185  fundamental 
x:  73  y:  15 ratio: 0.205479  fundamental 
x:  111  y:  23 ratio: 0.207207  fundamental 
x:  801  y:  167 ratio: 0.208489
x:  1223  y:  255 ratio: 0.208504
x:  3477  y:  725 ratio: 0.208513
x:  5309  y:  1107 ratio: 0.208514
x:  38429  y:  8013 ratio: 0.208514
x:  58677  y:  12235 ratio: 0.208514
x:  166823  y:  34785 ratio: 0.208514
x:  254721  y:  53113 ratio: 0.208514
x:  1843791  y:  384457 ratio: 0.208514
x:  2815273  y:  587025 ratio: 0.208514
x:  8004027  y:  1668955 ratio: 0.208514
x:  12221299  y:  2548317 ratio: 0.208514

Thu Mar 31 11:00:14 PDT 2016

 x^2 - 23 y^2 = 154

jagy@phobeusjunior:~$

I decided to draw the complete diagram of the Conway topograph, first the river, then the two extensions (trees) away from the river that, together, give all orbits for representing $x^2 - 23 y^2 = 154,$ those four seed pairs being $$ (19,3); \; (27,5); \; (73,15); \; (111,23). $$

As far as the original posted problem, the seeds for representing $x^2 - 23 y^2 = -11$ all occur along the river. Note that, as $x^2 - 23 y^2 \equiv x^2 + y^2 \pmod 4,$ whenever $x^2 - 23 y^2 \equiv 0 \pmod 4,$ it follows that both $x,y$ are even. That is, the seeds for $-11$ are $$ (9,2); \; (14,3), $$ therefore the only seeds for $-44$ are $$ (18,4); \; (28,6). $$

You can see Ross's formula $(24x+115y, 5x+24y)$ at the far right of the river diagram, on graph paper. We see a representation of $1$ with column vector $(24,5)^T,$ below it and all the way to the edge of the paper, a representation of $-23$ with column vector $(115,24)^T.$ Put them side by side and we get the two by two matrix $$ \left( \begin{array}{cc} 24 & 115 \\ 5 & 24 \end{array} \right) $$ of determinant $+1.$ That matrix, applied to a column vector $(x,y)^T,$ gives Ross's mapping.

I used a pink pen for the represented numbers in the two tree diagrams, it is a bit hard to read; next time, always red for the represented numbers.

I put four explanatory documents at OTHER with prefix indefinite_binary. For that matter, Conway's entire book is available at PDF

enter image description here

enter image description here

enter image description here

| cite | improve this answer | |
$\endgroup$

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.