Suppose $x,y,z,w \in \mathbb{F}_{q^2}$, where $q=p^k$ for some prime $p$. Consider the system of equations $$ \left\{ \begin{array}{l} xy + zw = 0; \\ xy^q + yx^q + zw^q + wz^q = 0. \end{array} \right. $$

Is it possible to say something about the number of solutions (excluding the trivial one) over $\mathbb{F}_{q^2}$? I am not that familiar with the finite field arithmetics, so sorry if this is something known.

If we substitute $xy$ from the first equation to the second, we get $$ (xy)(x^{q-1}+y^{q-1}-z^{q-1}-w^{q-1}) = 0, $$ but this does not make it easier for me. Thanks in advance for your help and suggestions!

UPD: By performing some computations in Mathematica, I have found that for $q=2,3,5,7$ the number of solutions takes respectively values $64,513,6625,35329$ which suggests that the number of solutions in $\mathbb{F}_{q^2}$ must be $q^3(2q^2+q-2)$, however I still don't know how to prove this in general.


To just get started, consider the fact that if $(a,b,c,d)$ is a solution, then so is $(\lambda a, \lambda b, \lambda c, \lambda d)$ for every $\lambda \in \mathbb{F}_{q^{2}}$. Use this to consider two cases: first, $x = 0$ (this is rather easy; I will leave this part to you). Second, if $x \neq 0$, then you can assume $x = 1$. At the end, you can multiply the number of solutions with $x = 1$ by $q^{2}-1$ to get the total number of solutions with $x \neq 0$.

Addressing the second case, taking $x = 1$ we then have $y = -zw$ in order for the first equation to hold. Then the second equation becomes $$-z^{q}w^{q} - zw +zw^{q}+z^{q}w=0.$$ This simplifies to $$(z-z^{q})(w^{q}-w) = 0,$$ implying that either $z = z^{q}$ or $w = w^{q}$. This happens when $z \in \mathbb{F}_{q}$ or $w \in \mathbb{F}_{q}$, respectively, so there are $2q^{3}-q^{2}$ such solutions.

Remember this only counts solutions with $x = 1$, so we have to multiply these by $(q^{2}-1)$ to get a total of $2q^{3}(q^{2}-1)$ solutions with $x \neq 0$. Then there are $2q^{4}-q^{2}$ further solution with $x=0$. This gives a total of $$(2q^{3}-q^{2})(q^{2}-1)+2q^{4}-q^{2} = q^{3}(2q^{2}+q-2)$$ solutions.

For a more technical description of this problem, you can think of your vectors $(x,y,z,w)$ as being points in a projective $3$-space $PG(3,q)$; this means that two vectors represent the same point iff they are a scalar multiple of each other (0 vector does not represent a point). There are $q^3+q^2+q+1$ points. Under this model you are looking for the intersection between a hyperbolic quadric (contains $(q^{2}+1)^{2}$ points) and a Hermitian surface (contains $(q^{2}+1)(q^{3}+1)$ points). Each point in this intersection then corresponds to $q^{2}-1$ nonzero vectors that satisfy both equations.

  • $\begingroup$ Can you suggest some literature? Actually, I am trying to consider the four different possibilities: 1) $xy = xy^q + x^qy = 0, zw = zw^q + z^qw$; 2) $xy = 0, xy^q + x^qy = a, zw = 0, zw^q + z^qw = -a$; 3) $xy = b, xy^q + x^qy = 0, zw = -b, zw^q + z^qw = 0$; 4) $xy = b, xy^q + x^qy = a, zw = -b, zw^q + z^qw = -a$. The last case is the hardest one, so I am having huge problems with it. $\endgroup$ – kissanpentu May 27 '16 at 23:50
  • $\begingroup$ Thank you very much for such a brilliant answer! However, could you still recommend some references for the counting techniques in projective spaces as I am trying to find an inductive technique to count the points in $2k$-dimensional space which satisfy both hermitian form and the quadratic form of hyperbolic type (by "satisfy" I mean being isotropic)? $\endgroup$ – kissanpentu May 28 '16 at 12:46
  • $\begingroup$ And tiny correction: when $x=0$ there are actually $2q^2-1$ solutions for the remaining system (not $2q^2-2$, since we only need to count the case $z=0,w=0$ once), so there are $2q^4-q^2$ solutions in total for this case. But anyway, thanks, you have saved the day for me :) $\endgroup$ – kissanpentu May 28 '16 at 13:38
  • $\begingroup$ @kissanpentu Maybe a good reference, it's extensive though. Feel free to skip ahead to what interests you. math.ucdenver.edu/~spayne/classnotes/topics.pdf $\endgroup$ – Morgan Rodgers May 29 '16 at 14:53

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.