Let $k$ be any field (maybe need to be algebraically closed), and $n$ be a positive integer, $k[x_1,\ldots,x_n]$ be the polynomial ring. Then consider the equation $$\sum_{i=1}^n l_i(x)x_i^2=q(x)(x_1+\ldots+ x_n)$$ where $l_i(x)\in k[x_1,\ldots,x_n]$ is a homogeneous polynomial of degree $1$, and $q(x)\in k[x_1,\ldots,x_n]$ is a homogeneous polynomial of degree $2$. We want to find the solution $(l_i,q)$ for the equation above. Easy to see $$l_i=x_1+\ldots+ x_n$$ $$q=x_1^2+\ldots+ x_n^2$$ is a trivial solution (and any $\mathbb C$ multiple, i.e. $c_i l_i$ for $c_i\in k$). My question is

is there any other non-trivial solution?

In general I am concerned about the solution to $$\sum_{i=1}^n l_i(x)x_i^2=q(x)(\text{a general linear equation})$$ but let's see the special case first. I have no idea how to do this.


Robert's answer shows there are other solutions for $n<5$. Now I want to know the case $n=5$.

  • $\begingroup$ For $n=2$, take $l_i(x)=x_i$ and $q=x_1^2-x_1x_2+x_2^2$ as an example. $\endgroup$ – Mohan Oct 14 '18 at 22:20
  • $\begingroup$ @Mohan Thanks for your example. But maybe I should require $n>2$ since this is too special. $\endgroup$ – User X Oct 14 '18 at 22:27

Algebraically closed is not an issue: this involves solving a linear system of equations for the coefficients of the $l_i$ and $q$, with integer coefficients.

You will find other nontrivial solutions. For example, for each $i$ you could have $l_i = x_1 + \ldots + x_n$, all other $l_j = 0$, $q(x) = x_i^2$.

In the case $n=3$ you could also take $l_1(x) = x_2 - x_3$, $l_2(x) = x_1$, $l_3(x) = -x_1$, $q(x) = x_1 x_2 - x_1 x_3$ (and similarly for permutations of the variables).

In the case $n=4$ you could take $$ \eqalign{l_{{1}}(x)&=x_{{2}}-x_{{4}}\cr l_{{2}}(x)&=x_{{1}}-x_{{3}}\cr l_{{3}}(x)&=-x_{{2}}+x_{{4}}\cr l_{{4}}(x)&=-x_{{1}}+x_{{3}}\cr q(x)&=x_{{1}}x_{{2}}-x_{{1}}x_{{4}}-x_{{2}}x_{{3}}+x_{{3}}x_{{4}}\cr}$$ (and again permutations of this).

I suspect this type of solution doesn't exist for $n\ge 5$.

  • $\begingroup$ Thanks for your answer! But in fact I am mostly interested in the case $n=5$. I guess there is no solution for $n=5$, and thought it is a general fact for all $n$ (now you shows it is not). So, is there a way to show no solution for $n=5$? $\endgroup$ – User X Oct 15 '18 at 6:35
  • $\begingroup$ And note that the first solution you provided $l_i = x_1 + \ldots + x_n$, all other $l_j = 0$, $q(x) = x_i^2$, should be regarded as an trivial solution (the multiple of $l_i$, coefficients vary for different $i$, sorry for did not make it clear) $\endgroup$ – User X Oct 15 '18 at 6:51
  • $\begingroup$ Ok I have proved there is no trivial solution for $n\geq 5$. $\endgroup$ – User X Oct 15 '18 at 19:05
  • $\begingroup$ You mean no nontrivial solution? $\endgroup$ – Robert Israel Oct 16 '18 at 3:04
  • $\begingroup$ Yeah, it is a typo :) The proof is just by assume a solution and compare coefficients on both sides. $\endgroup$ – User X Oct 16 '18 at 21:18

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.