Is my proof that $(x,y)\subset \mathbb Q[x,y]$ is not a principal ideal correct?

Suppose $(x,y)=(f(x,y))$. Since $x,y\in (x,y)$, the degree of $f$ in $x$ and $y$ (separately) cannot exceed 1. So $f(x,y)=ax+by+c$. Since $x\in (x,y)$, there must exist $g(x,y)$ such that $f(x,y)g(x,y)=x$. The only event in which this happens is when $f(x,y)=ax,\ g(x,y)=1/a$. So $b=c=0$. Since $y\in (x,y)$, $a=c=0$. Thus $f(x,y)=0$. But $(x,y)$ is not the zero ideal, a contradiction.

  • $\begingroup$ It's quite fine for me. $\endgroup$
    – Bernard
    Nov 1, 2018 at 23:28

3 Answers 3


I second Alex J Best's comment about a little proof fix, but also I think it's worth noting that this holds in $R[x,y]$ for $R$ any commutative ring with identity. So $\mathbb{Q}$ is not special in this regard. Your proof will continue to work for any integral domain. But if $R$ has zero divisors, you cannot immediately say that $f$ has degree $1$, and the proof is a little more complicated.

You might like to try to prove the following for a general ring $R$

$f \in R[x,y]$ divides $x$ and $y$ iff $f$ is a unit

This immediately implies that $(x,y)$ is not principal. I include a proof sketch in the spoiler below if you are interested. Like your proof, it mostly involves comparing coefficients.

Proof: Write $fg = x$ and $fh = y$. Let $f_x, f_y$ and $f_0$ denotethe $x$ coefficient, $y$ coefficient, and constant coefficient of $f$. First we can show that $f_0$ is a unit. Suppose not the case. Looking at $1 = f_0g_x + f_xg_0$ and $0 = f_0 g_y + f_y g_0$ modulo $(f_0)$, we deduce $f_x \notin (f_0), f_y \in (f_0)$. But similarly we could do the same with $fh = y$ to deduce $f_y \notin (f_0), f_x \in (f_0)$. Together these yield a contradiction, so conclude $f_0$ is a unit.

Now, knowing that $f_0$ is a unit, compare coefficients in $fg = x$ to see that $g \in R[x]$, and similarly $h \in R[y]$. Considering $fg = 0 \text{ mod } x$ we see that $x$ divides $g$, and similarly $y$ divides $h$, which gives us the equation $yg = xh$. Comparing coefficients in $yg = xh$ we finally see $g = x$ and $h = y$ up to unit multiples. It follows that $f$ is a unit.

  • $\begingroup$ If $R=F$ is a field, is your lemma obvious? In that case, $F[x,y]$ is a UFD, so $x$ is irreducible iff $x$ is prime, and $x$ is certainly prime because $F[x,y]/(x)$ is the domain $F[y]$. Similarly, $y$ is irreducible. Moreover, they ($x$ and $y$) are not associates because they generate different ideals. So if $f$ divides the two non-associate irreducible elements $x,y$, then $f$ must be a unit. $\endgroup$
    – user557
    Nov 2, 2018 at 1:28
  • $\begingroup$ The proof you just gave works perfectly so long as $x$ and $y$ are principal primes, which is the case exactly when $R = D$ is a domain. $\endgroup$ Nov 2, 2018 at 1:46
  • $\begingroup$ What do you mean by "principal primes"? I think my proof uses that irreducible elements in $R[x,y]$ are prime, which holds in UFDs. But maybe there is a different way to prove that $x,y$ are irreducible in $R[x,y]$ when $R$ is a domain... $\endgroup$
    – user557
    Nov 2, 2018 at 1:46
  • $\begingroup$ I just mean that the elements $x$ and $y$ are prime (the ideals they generate are prime), sorry for the confusion. Like you mentioned, you know that $(x)$ is prime because $D[x,y]/(x) = D[y]$ is a domain. Prime elements are always irreducible. $\endgroup$ Nov 2, 2018 at 1:51
  • $\begingroup$ Oh, right, I don't need that irreducible implies prime. $\endgroup$
    – user557
    Nov 2, 2018 at 1:53

Its pretty much there, but I think a little more is needed, for example:

"The only event in which this happens is when $f(x,y)=ax, g(x,y)=1/a$"

we could have the other way around also right? For example $f(x,y) = 1$, $g(x,y) = x$. You should say also why this can't happen!


Your proof is almost fine, it just needs a couple of details. But you're overcomplicating things.

Consider $f(x,y)g(x,y)=x$ as an equality between polynomials in $y$ over $\mathbb{Q}[x]$. As this has degree $0$, both $f$ and $g$ must have degree $0$. Thus $b=0$. Similarly, $a=0$.

Therefore $f(x,y)=c$, which either generates the whole ring or the zero ideal. On the other hand, $(x,y)$ is a nonzero proper ideal, as $\mathbb{Q}[x,y]/(x,y)\cong\mathbb{Q}$.

  • $\begingroup$ What do you mean by "identity between polynomials in $y$ over $\mathbb Q[x]$"? $\endgroup$
    – user557
    Nov 3, 2018 at 2:13
  • $\begingroup$ @user437309, $\mathbb Q[x,y]\cong (\mathbb Q[x])[y]$, thus $f$ anf $g$ can be thought of as polynomials in single variable $y$ with coefficients in $\mathbb Q[x]$. $\endgroup$
    – Ennar
    Nov 3, 2018 at 2:22
  • $\begingroup$ Ohh, for some reason I interpreted "identity" as an identity in some ring, and not as an "equality". $\endgroup$
    – user557
    Nov 3, 2018 at 3:16
  • $\begingroup$ @user437309 Changed the word $\endgroup$
    – egreg
    Nov 3, 2018 at 10:11

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.