Find integer points on this equation

  • Thread starter Thread starter littlemathquark
  • Start date Start date
  • Tags Tags
    Integer
AI Thread Summary
The discussion revolves around finding integer solutions for the equation x^2 - xy - 6x + y^2 + 5y + 6 = 0. Several integer points have been identified, including (0, -2), (0, -3), (3, 1), (3, -3), (4, -2), and (4, 1). Participants debate the validity of certain transformations and the completeness of the solution set, referencing tools like Wolfram Alpha for verification. The conversation also touches on the application of theorems related to sums of squares to determine the existence of additional solutions. Ultimately, the confirmed integer solutions are reiterated, emphasizing the need for careful calculation and theoretical understanding.
littlemathquark
Messages
204
Reaction score
26
Homework Statement
Find integer points on ##x^2-xy-6x+y^2+5y+6=0##
Relevant Equations
Find integer points on ##x^2-xy-6x+y^2+5y+6=0##
One of my solution is:
##x^2-xy-6x+y^2+5y+6=(x-y)^2+(x-6)^2+(y+5)^2=49## and by trying integer ##(x,y)## points are ##(0,-2),(0,-3),(3,1),(3,-3),(4,-2),(4,1)##

Another idea ##y^2+(5-x)y+x^2-6x+6=0## or ##x^2-(y+6)x+y^2+5y+6=0## and solving quadratic equation considering to integer solutions.

My question is: Are there other solutions that I can't see and is there a criterion to (number of ) find integer-coordinate points on the curve defined by the given equation?
 
Last edited:
Physics news on Phys.org
littlemathquark said:
One of my solution is: ##x^2-xy-6x+y^2+5y+6=(x-y)^2+(x-6)^2+(y+5)^2=49=0## and by trying integer ##(x,y)## points are ##(0,-2),(0,-3),(3,1),(3,-3),(4,-2),(4,1)##
You can't get from ##x^2 - xy \dots## to ##(x - y)^2 \dots##.
Also, the part ##\dots = 49 = 0## is almost certainly a typo in which you hit = instead of +.
 
Last edited:
Ignoring the fact that you lost a ##xy## term in your "equation", let's consult WA:
https://www.wolframalpha.com/input?i=x^2-xy-6x+y^2+5y+6=0

It not only lists the integer solutions, it also shows how to prove that there aren't others by inspecting the root expressions.
 
fresh_42 said:
Ignoring the fact that you lost a ##xy## term in your "equation", let's consult WA:
https://www.wolframalpha.com/input?i=x^2-xy-6x+y^2+5y+6=0

It not only lists the integer solutions, it also shows how to prove that there aren't others by inspecting the root expressions.
But why the solution ##(4,1)## is not in the solution list?
 
Mark44 said:
You can't get from ##x^2 - xy \dots## to ##(x - y)^2 \dots##.
Also, the part ##\dots = 49 = 0## is almost certainly a typo in which you hit = instead of +.
##2(x^2-xy-6x+y^2+5y+6)=0*2##
##(x-y)^2+x^2-12x+y^2+10y+12=0##
 
Last edited by a moderator:
littlemathquark said:
But why the solution ##(4,1)## is not in the solution list?
I guess they only display five solutions. It is not difficult to complete the calculations:
\begin{align*}
a^2&=-3x^2+14x+1\, , \,2y=\pm a+x-5\\
3x&=7\pm \sqrt{52-3a^2}\\
3x&\in 7\pm \{2,5,7\}=\{0,2,5,9,12,14\}\\
(x,a)&\in \{(0,\pm 1),(3,\pm 4),(4,\pm 3)\} \\
(x,y)&\in \{(0,-2),(0,-3),(3,1),(3,-3),(4,-2),(4,1)\}
\end{align*}
It needs a little bit of effort from your side.

And if you click on "More Solutions" you will find all six solutions.
 
  • Like
Likes chwala, BvU and littlemathquark
I also considered using Pick's theorem, but I abandoned the idea because I didn't know the number of lattice/integer points inside the ellipse. Actually, I'm looking for a more general theory that gives the number of lattice points on a curve.
 
littlemathquark said:
I also considered using Pick's theorem, but I abandoned the idea because I didn't know the number of lattice/integer points inside the ellipse. Actually, I'm looking for a more general theory that gives the number of lattice points on a curve.
Here are some papers I found by searching "lattice points on a curve + pdf"
https://math24.wordpress.com/wp-content/uploads/2013/02/lattice-point.pdf
https://www.worldscientific.com/doi/abs/10.1142/S1793042124500994
https://core.ac.uk/download/pdf/82591791.pdf

I doubt that there is one-solution-fits-all.
 
  • Like
Likes littlemathquark
Another solution. I used "Fermat sum of two squares theorem". What about solution?

The equation can be written as ##(x-y)^2+(x-6)^2+(y+5)^2=49## by multiplying main equation by 2. From this equation, it is clear that ##y\le1##.
Theorem (Fermat's theorem on sums of two squares): A prime number p > 2 is the sum of two squares if and only if p ≡ 1 (mod 4).
For ##y = 1##, the right side of the equation ( I mean ##(x-1)^2+(x-6)^2=49-36=13##) is p = 13 = 4k + 1, so the equation has solutions. By trial and error, these solutions are found to be x = 3 and x = 4.

Theorem (theorem on sums of two squares): For a composite number containing a prime factor of the form 4k + 3 to be expressible as the sum of two squares, the exponents of these prime factors must be even. Only composite numbers containing prime factors of the form 4k + 1 can be written as the sum of two squares.

Now, let's continue using this theorem:

For ##y = 0##, the right side of the equation is ##24 = 2^3 · 3##, and since the exponent of the factor 3, which is of the form 4k + 3, is not even, it cannot be written as the sum of two squares. So, the equation has no solution.

For ##y = -1##, the right side is 33 = 3 · 11, and for the same reason, 33 cannot be written as the sum of two squares.

For ##y = -2##, the right side is ##40 = 2^3 · 5##, and since 5 is of the form 4k + 1, 40 can be written as the sum of two squares. By trial and error, x = 0 and x = 34 are found.

For ##y = -3##, the right side is ##45 = 2^3 · 5##, and for the same reason, 45 can be written as the sum of two squares. By trial and error, x = 0 and x = 3 are found.

For ##y = -4##, the equation ##(x+4)^2+(x-6)^2=48## does not represent a real circle.

Accordingly, all solutions are ##(3, 1), (4, 1), (0, -2), (4, -2), (0, -3), (3, -3)##.
 
Last edited:
Back
Top