MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

i am wondering if there is a complete solution for the equation $a^2+pb^2-2c^2-2kcd+(p+k^2)d^2=0$ in which $a,b,c,d,k$ are integer(not all zero) and $p$ is odd prime.

share|cite|improve this question
Evidently, there are solutions when $p=3$.^2%2B3*b^2%E2%88%922*c^2%E2%88‌​%922*kcd%2B%283%2Bk^2%29*d^2%3D0 – Charles Siegel Sep 10 '10 at 21:20
Extremely unlikely that there is a parametrization of all solutions. What is your background in integral quadratic forms, and, as usual, why do you want to know this? – Will Jagy Sep 10 '10 at 21:41

I'm not getting much 2-adic information for this one, but it should be easy enough to check all solutions mod 8 and mod 16 and see what happens.

To restrict anything, one property requires $p \equiv \pm 3 \pmod 8$ and the other requires $p \equiv 3 \pmod 4.$ Put them together, when $$p \equiv 3 \pmod 8 $$ and $$ p | k, $$ then all four of your letters $$ a,b,c,d = 0.$$ The proof uses two flavors of anisotropy for binaries. The assumption is that at least one of $ a,b,c,d $ is nonzero and $\gcd(a,b,c,d) = 1.$ First we have forced $a^2 - 2 c^2 \equiv 0 \pmod p,$ so $a,c \equiv 0 \pmod p$ as $(2 | p) = -1.$ But then $ p b^2 + p d^2 \equiv 0 \pmod {p^2},$ or $ b^2 + d^2 \equiv 0 \pmod p,$ so $b,d \equiv 0 \pmod p$ as $(-1 | p) = -1.$ So $ p | \gcd(a,b,c,d)$ contrary to assumption.

Otherwise, given a fixed $(p,k)$ once you have a nontrivial solution you get infinitely many using automorphs of the indefinite part in variables $(c,d).$ That is, there may be many parametrized families of solutions of one type or another. But you can figure some of those out with a computer algebra system more easily than I can by hand.

The next interesting case is when $12 k^2 + 8p$ is a square, which means that the binary form $T(c,d)=2c^2+2kcd-(p+k^2)d^2$ factors. So $3 k^2 + 2p$ is a square, which is not possible for even $k,$ so $k$ is odd and $2p \equiv 6 \pmod 8,$ or $p \equiv -1 \pmod 4.$ Unless $p=3$ we also need $p \equiv -1 \pmod 3,$ or $p \equiv -1 \equiv 11 \pmod {12}.$

For example, with $p=11, k=1,3 k^2 + 2p = 25, p + k^2 = 12,$ we have $$ a^2+11b^2-2c^2-2cd+12d^2 = a^2+11b^2-2(c-2d)(c+3d).$$ The value of the factorization is that we can take, for instance, $c = 2 d + 1, c + 3 d = 5 d + 1,$ and $$ a^2+11b^2-2(5d+1) = 0.$$ Now $a^2 + 11 b^2$ is not even unless it is also divisible by $4.$ We also need $ a^2 \equiv b^2 \equiv 1 \pmod 5.$ Put them together, we have a parametrized solution of sorts, with $$ a \equiv 1,4 \pmod 5, \; \; b \equiv 1,4 \pmod 5, \; \; a \equiv b \pmod 2$$ take $c = 2 d + 1$ and $$ d = \frac{ a^2+11b^2-2}{10}.$$

share|cite|improve this answer

Perhaps it would help if we knew where the question is coming from.

For what it's worth, you can write the equation in the form $$ a^2 + (c-kd)^2 + pb^2 + pd^2 = 3c^2, $$ so you are looking at a parametrized subset of the equation $$ A^2 + B^2 + pC^2 + pD^2 = 3c^2. $$ If $p \equiv 1 \bmod 4$, then $p(B^2+D^2) = R^2 + S^2$ is a sum of two squares, and your solutions must occur among those of $$ R^2 + S^2 + T^2 + U^2 = 3c^2. $$ Both quadrics can be parametrized by the standard method of sweeping lines if you know one solution. For arbitrary primes $p$ such a solution seems to be difficult to find. And even armed with such a parametrization you then would have to figure out which of them satisfy the additional conditions coming from the original equation.

share|cite|improve this answer

There are lots! Here is a selection:




share|cite|improve this answer
thank you but i need complete solution – M.S Sep 10 '10 at 21:25
If $k=0$ and $p \equiv 3 \pmod 8$ then all the other variables are $0$ also. – Will Jagy Sep 10 '10 at 21:32

For the equation:


If the number $k$ is the problem any, and $p$ is such as this: $p=\frac{t^2}{2}-1$

Then the solution can be written:





$n,s,j,r$ - integers which we are set.

If you can represent numbers as: $p=3k^2-t^2$

This decision when the coefficients are related through the equation of Pell. $t^2-3k^2=-p$

To simplify calculations we will make this change.





Then the solution can be written:





$n,s,e,j$ - integers which we ask.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.