TheoremComplete

Bézout's Theorem

Bézout's theorem is the fundamental intersection-counting result in projective geometry. It tells us that two projective plane curves of degrees dd and ee always meet in exactly dede points — provided we count correctly.


Statement

Theorem1.2Bézout's Theorem

Let C=V(F)C = V(F) and D=V(G)D = V(G) be projective plane curves in P2\mathbb{P}^2 over an algebraically closed field kk, where degF=d\deg F = d and degG=e\deg G = e. If CC and DD have no common component (i.e., gcd(F,G)=1\gcd(F, G) = 1), then

PCDI(P,CD)=de\sum_{P \in C \cap D} I(P, C \cap D) = de

where I(P,CD)I(P, C \cap D) is the intersection multiplicity of CC and DD at PP.

Definition1.13Intersection multiplicity

For two curves C=V(F)C = V(F) and D=V(G)D = V(G) in P2\mathbb{P}^2 meeting at a point PP, the intersection multiplicity is

I(P,CD)=dimkOP,P2/(F,G)I(P, C \cap D) = \dim_k \mathcal{O}_{P,\mathbb{P}^2} / (F, G)

where OP,P2\mathcal{O}_{P,\mathbb{P}^2} is the local ring at PP (in an affine chart containing PP). Intuitively, this counts "how many times" CC and DD meet at PP.

I(P,CD)=1I(P, C \cap D) = 1 iff CC and DD are transverse at PP (different tangent lines).


Examples: Bézout in action

ExampleTwo lines (d=1, e=1): 1·1 = 1

Two distinct lines in P2\mathbb{P}^2 meet in exactly one point. For example, V(x)V(x) and V(y)V(y) meet at [0:0:1][0:0:1]. This fails in A2\mathbb{A}^2 (parallel lines!), but in P2\mathbb{P}^2 every pair of lines intersects.

If the lines are V(x)V(x) and V(x+y)V(x+y), they meet at [0:0:1][0:0:1] with I=1I = 1 (transverse).

ExampleLine meets conic (1·2 = 2)

The line V(y)V(y) and the conic V(x2+y2z2)V(x^2 + y^2 - z^2) in P2\mathbb{P}^2:

Setting y=0y = 0: x2=z2x^2 = z^2, so [1:0:1][1:0:1] and [1:0:1][1:0:-1] (or equivalently [1:0:1][-1:0:1]). Two intersection points, each with multiplicity 1. Total: 12=21 \cdot 2 = 2. ✓

ExampleTangent line to a conic (multiplicity 2)

The line V(yz)V(y - z) and the parabola V(yzx2)V(y z - x^2) (homogenization of y=x2y = x^2):

Setting y=zy = z: z2=x2z^2 = x^2, so [1:1:1][1:1:1] and [1:1:1][-1:1:1]. Wait — that's still 2 distinct points, each with I=1I = 1.

Now try V(z)V(z) (the line at infinity) and V(yzx2)V(yz - x^2): setting z=0z = 0, x2=0x^2 = 0, so only [0:1:0][0:1:0] with multiplicity 2. The line at infinity is tangent to the parabola. Total: 12=21 \cdot 2 = 2. ✓

ExampleTwo conics (2·2 = 4)

V(x2+y2z2)V(x^2 + y^2 - z^2) and V(x2y2z2)V(x^2 - y^2 - z^2) in P2\mathbb{P}^2:

Adding: 2x22z2=02x^2 - 2z^2 = 0, so x=±zx = \pm z. Substituting x=zx = z: z2+y2z2=y2=0z^2 + y^2 - z^2 = y^2 = 0, giving [1:0:1][1:0:1]. Substituting x=zx = -z: similarly [1:0:1][1:0:-1].

Wait — only 2 points? Each must have multiplicity 2. Indeed, at [1:0:1][1:0:1], locally set z=1z = 1: curves are x2+y21x^2 + y^2 - 1 and x2y21x^2 - y^2 - 1. At (1,0)(1, 0): the Jacobian rows are (2,0)(2, 0) and (2,0)(2, 0) — same tangent direction! So I=2I = 2 at each point. Total: 2+2=4=222 + 2 = 4 = 2 \cdot 2. ✓

ExampleLine meets cubic (1·3 = 3)

A generic line meets a smooth cubic curve in 3 points. For the cubic V(y2zx3+xz2)V(y^2 z - x^3 + xz^2) (an elliptic curve), the line V(y)V(y) gives:

0=x3+xz2=x(z2x2)=x(zx)(z+x)0 = -x^3 + xz^2 = x(z^2 - x^2) = x(z-x)(z+x)

Three points: [0:0:1][0:0:1], [1:0:1][1:0:1], [1:0:1][-1:0:1], each with I=1I = 1. Total: 3=133 = 1 \cdot 3. ✓

This is the geometric basis for the group law on elliptic curves: three collinear points sum to zero.

ExampleFlex tangent to a cubic (multiplicity 3)

The inflection point of a cubic: at a flex point PP, the tangent line meets the curve with multiplicity 3 (all of the "budget" 13=31 \cdot 3 = 3 is concentrated at one point).

For V(y2zx3)V(y^2z - x^3) (the cuspidal cubic), the tangent at the cusp [0:0:1][0:0:1] is V(y)V(y). Setting y=0y = 0: 0=x30 = -x^3, so [0:0:1][0:0:1] with multiplicity 3. This is an extreme case where the tangent line has maximal contact.

ExampleTwo cubics (3·3 = 9) and the Cayley–Bacharach theorem

Two general cubics in P2\mathbb{P}^2 meet in 33=93 \cdot 3 = 9 points. The Cayley–Bacharach theorem states:

If two cubics C1,C2C_1, C_2 meet in 9 points, and a third cubic C3C_3 passes through 8 of them, then C3C_3 also passes through the 9th.

This is the geometric heart of the group law on elliptic curves: if EE is a smooth cubic (an elliptic curve) and L1,L2L_1, L_2 are two lines, then V(L1L2)=V(L1)V(L2)V(L_1 L_2) = V(L_1) \cup V(L_2) is a (degenerate) cubic meeting EE in 6 points. Cayley–Bacharach applied to these 6 points (via a suitable third cubic) yields the associativity of the group law.

ExampleWhy Bézout needs projective space

In A2\mathbb{A}^2: the circle x2+y2=1x^2 + y^2 = 1 and the line x=2x = 2.

4+y2=1y2=34 + y^2 = 1 \Rightarrow y^2 = -3: no real solutions! Over C\mathbb{C}: y=±i3y = \pm i\sqrt{3}, so 2 intersection points. Bézout gives 12=21 \cdot 2 = 2. ✓ (over C\mathbb{C}).

Now consider x2+y2=1x^2 + y^2 = 1 and x=0x = 0 (the yy-axis). Two intersection points (0,±1)(0, \pm 1). All accounted for.

But x2+y2=1x^2 + y^2 = 1 and the "line at infinity": homogenize to X2+Y2Z2X^2 + Y^2 - Z^2 and Z=0Z = 0. Setting Z=0Z = 0: X2+Y2=0X^2 + Y^2 = 0, giving [1:i:0][1:i:0] and [1:i:0][1:-i:0] over C\mathbb{C}. These are the circular points at infinity — they lie on every circle!

ExampleBézout and the genus formula

For a smooth projective curve CP2C \subseteq \mathbb{P}^2 of degree dd, the genus is

g=(d1)(d2)2.g = \frac{(d-1)(d-2)}{2}.

This follows from Bézout + the adjunction formula. Examples:

| Degree dd | Genus gg | Type | |---|---|---| | 1 | 0 | Line (P1\cong \mathbb{P}^1) | | 2 | 0 | Conic (P1\cong \mathbb{P}^1) | | 3 | 1 | Cubic (elliptic curve) | | 4 | 3 | Quartic | | 5 | 6 | Quintic | | dd | (d1)(d2)2\frac{(d-1)(d-2)}{2} | General |

The genus measures the "complexity" of the curve. Curves of genus 0 are rational (birational to P1\mathbb{P}^1). Curves of genus 1 are elliptic. Curves of genus 2\geq 2 have finite automorphism groups (Hurwitz).


Higher-dimensional Bézout

Theorem1.2'Bézout in Pⁿ

In Pn\mathbb{P}^n, if hypersurfaces V(F1),,V(Fn)V(F_1), \ldots, V(F_n) of degrees d1,,dnd_1, \ldots, d_n meet in finitely many points, then (counted with multiplicity):

PI(P)=d1d2dn.\sum_P I(P) = d_1 \cdot d_2 \cdots d_n.

ExampleThree quadrics in P³

Three generic quadric surfaces in P3\mathbb{P}^3 meet in 222=82 \cdot 2 \cdot 2 = 8 points. For example:

V(x02x12),V(x02x22),V(x02x32)V(x_0^2 - x_1^2), \quad V(x_0^2 - x_2^2), \quad V(x_0^2 - x_3^2)

give x0=±x1=±x2=±x3x_0 = \pm x_1 = \pm x_2 = \pm x_3, i.e., 8 points [±1:±1:±1:±1][\pm 1 : \pm 1 : \pm 1 : \pm 1] (up to overall scaling: [1:±1:±1:±1][1:\pm 1:\pm 1:\pm 1], which is 8 points). Total: 8. ✓

RemarkBeyond Bézout

Bézout's theorem generalizes vastly:

  • Intersection theory on general varieties: the intersection product in the Chow ring A(X)A^*(X).
  • Hilbert polynomials compute intersection numbers via χ\chi.
  • Schubert calculus on Grassmannians: "How many lines meet 4 general lines in P3\mathbb{P}^3?" Answer: 2 (by Bézout-style computation on G(2,4)\mathbb{G}(2,4)).