16
$\begingroup$

Bezout's Theorem states that for two homogeneous polynomials $f(x,y,z), g(x,y,z)$ over an algebraically closed field of degrees $m,n$ respectively, such that the two polynomials do not share a common component, then the number of intersections of $f,g$ is equal to $mn$ counting multiplicity. Is there an analogue of this theorem for WEIGHTED homogeneous polynomials? That is, suppose that $w_1, w_2, w_3$ are three coprime positive integers, and for a given polynomial $h(x,y,z)$ let $e_1(h), e_2(h), e_3(h)$ denote the degrees of $x,y,z$ in $h$ respectively. We say that $h$ is weighted homogeneous of degree $d$ with weight $(w_1, w_2, w_3)$ if $h$ satisfies $w_1 e_1(h) + w_2 e_2(h) + w_3 e_3(h) = d$. If we allow $d$ to vary across all positive integers, then the resulting set of polynomials is the set of weighted homogeneous polynomials with weight $(w_1, w_2, w_3)$.

So my question is, is there an analogue to Bezout's Theorem in this setting? That is, is there a constant $W = W(w_1, w_2, w_3)$ which depends on $w_1, w_2, w_3$ such that if $f,g$ are two weighted homogeneous polynomials with weight $(w_1, w_2, w_3)$ with no common componets, then the number of intersections of $f,g$ is bounded by $W(w_1, w_2, w_3) \deg(f) \deg(g)$?

$\endgroup$
1
  • $\begingroup$ I would try to estimate, for large $d$, the dimension of the space of homogeneous polynomials of degree $d$ modulo multiples of $f$ and multiples of $g$. You could show that this did not depend on $f$ and $g$ other than their degrees, then find the correct answer for polynomials of the form $e_i^k$. This seems like it would work. I'll think about it and come back to it later. $\endgroup$
    – Will Sawin
    Nov 14, 2011 at 21:54

1 Answer 1

21
$\begingroup$

The classical Bézout theorem works for curves in the projective space $\mathbb{P}^2$.

In the case of weighted homogeneous polynomials one needs a Bézout theorem in the weighted projective plane $\mathbb{P}^2(w_1, w_2, w_3)$. Such a result can be found, for instance, in the paper by Bartolo, Martin-Morales and Ortigas-Galindo Q-resolutions and intersection numbers, Section 5. It turns out that the intersection number of two curves of equation $f=0$ and $g=0$ in $\mathbb{P}^2(w_1, w_2, w_3)$ is given by $$\frac{1}{w_1 w_2 w_3} \deg_{\omega}(f) \deg_{\omega}(g),$$ where $\deg_{\omega}$ denotes the weighted degree, see also auniket's comment below.

Since $\mathbb{P}^2(w_1, w_2, w_3)$ is a singular variety (with cyclic quotient singularities, hence $\mathbb{Q}$-factorial), this formula makes sense only as an intersection formula for $\mathbb{Q}$-divisors. In fact, if the zero locus of your polynomials intersect the singular locus of the weighted projective plane, it may happen that the corresponding Weil divisors are not Cartier. Consequently, one can obtain a rational intersection number, insted of an integer one.

For instance, let us consider $\mathbb{P}(1,1,2)$, which is isomorphic to a quadric cone in $\mathbb{P}^3$. If $x, y, z$ are the weighted homogeneous coordinates, the ruling of the cone is generated by $x=0$ and $y=0$; furthermore, the intersection numbers for any two curves $L_1$, $L_2 \subset \mathbb{P}(1,1,2)$ of equation $$\lambda_1x+\mu_1y=0, \quad \lambda_2x+\mu_2y=0 \quad (\lambda_i, \mu_i \in \mathbb{C})$$ is equal to $\frac{1 \cdot 1}{ 1 \cdot 1 \cdot 2} = \frac{1}{2}$.

This happens because a line $L$ in the ruling is not a Cartier divisor (since it passes through the vertex $[0:0:1]$) but $2L$ is Cartier, being linearly equivalent to a conic, i.e. to a hyperplane section of the cone. Now two hyperplane sections intersect in two points, so we have $(2L_1)(2L_2)=2$, that is $L_1L_2=\frac{1}{2}$.

$\endgroup$
2
  • 1
    $\begingroup$ Let me explicitly state, just for clarity, that $W(w_1, w_2, w_3) = \frac{1}{w_1w_2w_3}$. Of course in the OP's formula $\deg(f)$ and $\deg(g)$ should be interpreted as weighted degrees. $\endgroup$
    – pinaki
    Nov 14, 2011 at 22:56
  • $\begingroup$ Dear auniket, thank you for your comment. I will put the value of $W$ in the answer. $\endgroup$ Nov 14, 2011 at 22:59

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.