TheoremComplete

Noether's Formula

Noether's formula is the fundamental identity relating the arithmetic genus of a smooth projective surface to its Chern numbers. It constrains the numerical invariants of all surfaces and serves as the starting point for the geography of surfaces -- the study of which pairs of Chern numbers can be realized by smooth projective surfaces.


Statement

Theorem5.7Noether's formula

Let XX be a smooth projective surface over an algebraically closed field. Then:

Ο‡(OX)=112(KX2+e(X)),\chi(\mathcal{O}_X) = \frac{1}{12}(K_X^2 + e(X)),

where Ο‡(OX)=1βˆ’q+pg\chi(\mathcal{O}_X) = 1 - q + p_g is the holomorphic Euler characteristic, KX2=c1(X)2K_X^2 = c_1(X)^2 is the self-intersection of the canonical divisor, and e(X)=c2(X)e(X) = c_2(X) is the topological Euler characteristic.

Equivalently, in terms of Chern numbers:

12Ο‡(OX)=c12+c2.12\chi(\mathcal{O}_X) = c_1^2 + c_2.

RemarkNotation and conventions

Here:

  • q=h1(OX)q = h^1(\mathcal{O}_X) is the irregularity.
  • pg=h2(OX)=h0(KX)p_g = h^2(\mathcal{O}_X) = h^0(K_X) is the geometric genus.
  • c1=c1(Ξ©X1)c_1 = c_1(\Omega_X^1) so c12=KX2c_1^2 = K_X^2 (the first Chern class of the cotangent bundle equals the canonical class).
  • c2=c2(Ξ©X1)=c2(TX)=e(X)c_2 = c_2(\Omega_X^1) = c_2(T_X) = e(X) (the second Chern class equals the topological Euler characteristic).
  • Over C\mathbb{C}, we have e(X)=βˆ‘i=04(βˆ’1)ibi(X)e(X) = \sum_{i=0}^{4} (-1)^i b_i(X) where bib_i are the Betti numbers.

Derivation from Hirzebruch--Riemann--Roch

Theorem5.8Hirzebruch--Riemann--Roch for surfaces

For a smooth projective surface XX and a vector bundle (or coherent sheaf) E\mathcal{E}:

Ο‡(E)=∫Xch⁑(E)β‹…td⁑(TX).\chi(\mathcal{E}) = \int_X \operatorname{ch}(\mathcal{E}) \cdot \operatorname{td}(T_X).

The Todd class of the tangent bundle is td⁑(TX)=1+12c1(TX)+112(c1(TX)2+c2(TX))\operatorname{td}(T_X) = 1 + \frac{1}{2}c_1(T_X) + \frac{1}{12}(c_1(T_X)^2 + c_2(T_X)).

RemarkDeriving Noether's formula

Setting E=OX\mathcal{E} = \mathcal{O}_X (the trivial line bundle, D=0D = 0):

ch⁑(OX)=1\operatorname{ch}(\mathcal{O}_X) = 1, so Ο‡(OX)=∫Xtd⁑(TX)\chi(\mathcal{O}_X) = \int_X \operatorname{td}(T_X).

The integral picks out the degree-22 component of the Todd class:

Ο‡(OX)=112(c1(TX)2+c2(TX)).\chi(\mathcal{O}_X) = \frac{1}{12}(c_1(T_X)^2 + c_2(T_X)).

Since c1(TX)=βˆ’KXc_1(T_X) = -K_X, we have c1(TX)2=KX2c_1(T_X)^2 = K_X^2, and c2(TX)=e(X)c_2(T_X) = e(X). This gives:

Ο‡(OX)=112(KX2+e(X)).\chi(\mathcal{O}_X) = \frac{1}{12}(K_X^2 + e(X)).

Noether's formula is simply HRR applied to the structure sheaf.


Over the complex numbers

RemarkHodge-theoretic form

Over C\mathbb{C}, the Hodge decomposition gives Hk(X,C)=⨁p+q=kHp,q(X)H^k(X, \mathbb{C}) = \bigoplus_{p+q=k} H^{p,q}(X) with hp,q=dim⁑Hp,qh^{p,q} = \dim H^{p,q}. The Betti numbers are bk=βˆ‘p+q=khp,qb_k = \sum_{p+q=k} h^{p,q}.

For a surface: b0=1b_0 = 1, b1=2qb_1 = 2q, b2=2pg+h1,1b_2 = 2p_g + h^{1,1}, b3=2qb_3 = 2q, b4=1b_4 = 1.

The topological Euler characteristic is:

e(X)=1βˆ’2q+(2pg+h1,1)βˆ’2q+1=2+2pgβˆ’4q+h1,1.e(X) = 1 - 2q + (2p_g + h^{1,1}) - 2q + 1 = 2 + 2p_g - 4q + h^{1,1}.

Substituting into Noether's formula and writing Ο‡(OX)=1βˆ’q+pg\chi(\mathcal{O}_X) = 1 - q + p_g:

12(1βˆ’q+pg)=KX2+2+2pgβˆ’4q+h1,1,12(1 - q + p_g) = K_X^2 + 2 + 2p_g - 4q + h^{1,1},

which simplifies to KX2=10βˆ’8q+10pgβˆ’h1,1K_X^2 = 10 - 8q + 10p_g - h^{1,1}, or equivalently:

h1,1=10Ο‡(OX)βˆ’KX2+2(1βˆ’q).h^{1,1} = 10\chi(\mathcal{O}_X) - K_X^2 + 2(1 - q).

This determines h1,1h^{1,1} from KX2K_X^2, qq, and pgp_g.


Verification: classical examples

ExampleThe projective plane

For X=P2X = \mathbb{P}^2: KP2=βˆ’3HK_{\mathbb{P}^2} = -3H, so K2=9K^2 = 9. The Betti numbers are b0=1,b1=0,b2=1,b3=0,b4=1b_0 = 1, b_1 = 0, b_2 = 1, b_3 = 0, b_4 = 1, giving e(P2)=3e(\mathbb{P}^2) = 3.

Noether: Ο‡(OP2)=112(9+3)=1212=1\chi(\mathcal{O}_{\mathbb{P}^2}) = \frac{1}{12}(9 + 3) = \frac{12}{12} = 1.

Direct computation: h0(O)=1h^0(\mathcal{O}) = 1, h1(O)=0h^1(\mathcal{O}) = 0, h2(O)=0h^2(\mathcal{O}) = 0, so Ο‡(O)=1\chi(\mathcal{O}) = 1.

ExampleThe quadric surface

For X=P1Γ—P1X = \mathbb{P}^1 \times \mathbb{P}^1: K=βˆ’2F1βˆ’2F2K = -2F_1 - 2F_2, so K2=2β‹…(βˆ’2)(βˆ’2)β‹…F1β‹…F2=8K^2 = 2 \cdot (-2)(-2) \cdot F_1 \cdot F_2 = 8. The Betti numbers are b0=1,b1=0,b2=2,b3=0,b4=1b_0 = 1, b_1 = 0, b_2 = 2, b_3 = 0, b_4 = 1, giving e=4e = 4.

Noether: Ο‡(O)=112(8+4)=1\chi(\mathcal{O}) = \frac{1}{12}(8 + 4) = 1.

Direct: q=0q = 0, pg=0p_g = 0, so Ο‡(O)=1\chi(\mathcal{O}) = 1.

ExampleK3 surfaces

A K3 surface has KX=0K_X = 0 (so K2=0K^2 = 0), q=0q = 0, and pg=1p_g = 1.

Noether: Ο‡(OX)=1βˆ’0+1=2=112(0+e(X))\chi(\mathcal{O}_X) = 1 - 0 + 1 = 2 = \frac{1}{12}(0 + e(X)), hence e(X)=24e(X) = 24.

Betti numbers: b0=1,b1=0,b2=22,b3=0,b4=1b_0 = 1, b_1 = 0, b_2 = 22, b_3 = 0, b_4 = 1. Check: e=1+22+1=24e = 1 + 22 + 1 = 24.

Hodge diamond: h0,0=1h^{0,0} = 1, h1,0=h0,1=0h^{1,0} = h^{0,1} = 0, h2,0=h0,2=1h^{2,0} = h^{0,2} = 1, h1,1=20h^{1,1} = 20.

From the formula: h1,1=10(2)βˆ’0+2(1βˆ’0)=22h^{1,1} = 10(2) - 0 + 2(1 - 0) = 22... let us recheck: h1,1=10Ο‡βˆ’K2+2(1βˆ’q)=20βˆ’0+2=22h^{1,1} = 10\chi - K^2 + 2(1 - q) = 20 - 0 + 2 = 22. But b2=h2,0+h1,1+h0,2=1+20+1=22b_2 = h^{2,0} + h^{1,1} + h^{0,2} = 1 + 20 + 1 = 22. So indeed h1,1=20h^{1,1} = 20.

ExampleAbelian surfaces

An abelian surface AA has KA=0K_A = 0 (so K2=0K^2 = 0), q=2q = 2, pg=1p_g = 1.

Noether: Ο‡(OA)=1βˆ’2+1=0=112(0+e(A))\chi(\mathcal{O}_A) = 1 - 2 + 1 = 0 = \frac{1}{12}(0 + e(A)), hence e(A)=0e(A) = 0.

Betti numbers: b0=1,b1=4,b2=6,b3=4,b4=1b_0 = 1, b_1 = 4, b_2 = 6, b_3 = 4, b_4 = 1. Check: e=1βˆ’4+6βˆ’4+1=0e = 1 - 4 + 6 - 4 + 1 = 0.

Hodge diamond: h0,0=1h^{0,0} = 1, h1,0=h0,1=2h^{1,0} = h^{0,1} = 2, h2,0=h0,2=1h^{2,0} = h^{0,2} = 1, h1,1=4h^{1,1} = 4.

ExampleEnriques surfaces

An Enriques surface has 2KX∼02K_X \sim 0 but KX≁0K_X \not\sim 0, q=0q = 0, pg=0p_g = 0.

K2=0K^2 = 0 (since 2K∼02K \sim 0 implies 4K2=(2K)2=04K^2 = (2K)^2 = 0).

Noether: Ο‡(O)=1βˆ’0+0=1=112(0+e)\chi(\mathcal{O}) = 1 - 0 + 0 = 1 = \frac{1}{12}(0 + e), hence e=12e = 12.

Betti numbers: b0=1,b1=0,b2=10,b3=0,b4=1b_0 = 1, b_1 = 0, b_2 = 10, b_3 = 0, b_4 = 1. Check: e=1+10+1=12e = 1 + 10 + 1 = 12.

ExampleSmooth hypersurfaces in P3

Let XdβŠ‚P3X_d \subset \mathbb{P}^3 be a smooth hypersurface of degree dd. Then KXd=(dβˆ’4)H∣XdK_{X_d} = (d - 4)H|_{X_d} by adjunction, so K2=d(dβˆ’4)2K^2 = d(d-4)^2.

The Euler characteristic of XdX_d is computed from the Chern classes of P3\mathbb{P}^3 restricted to XdX_d. Using the exact sequence 0β†’TXdβ†’TP3∣Xdβ†’NXd/P3β†’00 \to T_{X_d} \to T_{\mathbb{P}^3}|_{X_d} \to N_{X_d/\mathbb{P}^3} \to 0:

e(Xd)=d3βˆ’4d2+6d.e(X_d) = d^3 - 4d^2 + 6d.

Noether: Ο‡(OXd)=112(d(dβˆ’4)2+d3βˆ’4d2+6d)=112(2d3βˆ’12d2+22d)=d(dβˆ’1)(dβˆ’2)6+1βˆ’d+d(dβˆ’1)2\chi(\mathcal{O}_{X_d}) = \frac{1}{12}(d(d-4)^2 + d^3 - 4d^2 + 6d) = \frac{1}{12}(2d^3 - 12d^2 + 22d) = \frac{d(d-1)(d-2)}{6} + 1 - d + \frac{d(d-1)}{2}... let us compute directly:

112(d(dβˆ’4)2+d3βˆ’4d2+6d)=112(d3βˆ’8d2+16d+d3βˆ’4d2+6d)=112(2d3βˆ’12d2+22d)=d6(d2βˆ’6d+11)\frac{1}{12}(d(d-4)^2 + d^3 - 4d^2 + 6d) = \frac{1}{12}(d^3 - 8d^2 + 16d + d^3 - 4d^2 + 6d) = \frac{1}{12}(2d^3 - 12d^2 + 22d) = \frac{d}{6}(d^2 - 6d + 11).

Specific values:

  • d=1d = 1 (plane): Ο‡=16(1βˆ’6+11)=1\chi = \frac{1}{6}(1 - 6 + 11) = 1, K2=9K^2 = 9, e=3e = 3.
  • d=2d = 2 (quadric): Ο‡=26(4βˆ’12+11)=1\chi = \frac{2}{6}(4 - 12 + 11) = 1, K2=8K^2 = 8, e=4e = 4.
  • d=3d = 3 (cubic): Ο‡=36(9βˆ’18+11)=1\chi = \frac{3}{6}(9 - 18 + 11) = 1, K2=3K^2 = 3, e=9e = 9.
  • d=4d = 4 (quartic = K3): Ο‡=46(16βˆ’24+11)=2\chi = \frac{4}{6}(16 - 24 + 11) = 2, K2=0K^2 = 0, e=24e = 24.
  • d=5d = 5 (quintic): Ο‡=56(25βˆ’30+11)=5\chi = \frac{5}{6}(25 - 30 + 11) = 5, K2=5K^2 = 5, e=55e = 55.
ExampleCubic surface

The cubic surface X3βŠ‚P3X_3 \subset \mathbb{P}^3 is rational: it is the blowup of P2\mathbb{P}^2 at 66 points. So q=0q = 0, pg=0p_g = 0, Ο‡(O)=1\chi(\mathcal{O}) = 1.

KX3=βˆ’H∣X3K_{X_3} = -H|_{X_3}, so K2=H2β‹…X3=3K^2 = H^2 \cdot X_3 = 3 (the degree). But K2=3K^2 = 3 for the cubic surface.

From the blowup description: KBl⁑6(P2)=βˆ’3H+E1+β‹―+E6K_{\operatorname{Bl}_6(\mathbb{P}^2)} = -3H + E_1 + \cdots + E_6, and K2=9βˆ’6=3K^2 = 9 - 6 = 3.

e=12Ο‡βˆ’K2=12βˆ’3=9e = 12\chi - K^2 = 12 - 3 = 9. Alternatively: b2=1+6=7b_2 = 1 + 6 = 7 (the class of the line plus 66 exceptional divisors), so e=1+7+1=9e = 1 + 7 + 1 = 9.

ExampleProducts of curves

Let X=C1Γ—C2X = C_1 \times C_2 where CiC_i has genus gig_i. Then:

  • KX=Ο€1βˆ—KC1+Ο€2βˆ—KC2K_X = \pi_1^* K_{C_1} + \pi_2^* K_{C_2}, so K2=2(2g1βˆ’2)(2g2βˆ’2)=8(g1βˆ’1)(g2βˆ’1)K^2 = 2(2g_1 - 2)(2g_2 - 2) = 8(g_1 - 1)(g_2 - 1).
  • e(X)=e(C1)β‹…e(C2)=(2βˆ’2g1)(2βˆ’2g2)=4(g1βˆ’1)(g2βˆ’1)e(X) = e(C_1) \cdot e(C_2) = (2 - 2g_1)(2 - 2g_2) = 4(g_1 - 1)(g_2 - 1).
  • q=g1+g2q = g_1 + g_2, pg=g1g2p_g = g_1 g_2.
  • Ο‡(OX)=1βˆ’(g1+g2)+g1g2=(1βˆ’g1)(1βˆ’g2)\chi(\mathcal{O}_X) = 1 - (g_1 + g_2) + g_1 g_2 = (1 - g_1)(1 - g_2).

Noether check: 112(8(g1βˆ’1)(g2βˆ’1)+4(g1βˆ’1)(g2βˆ’1))=12(g1βˆ’1)(g2βˆ’1)12=(g1βˆ’1)(g2βˆ’1)=(1βˆ’g1)(1βˆ’g2)\frac{1}{12}(8(g_1-1)(g_2-1) + 4(g_1-1)(g_2-1)) = \frac{12(g_1-1)(g_2-1)}{12} = (g_1-1)(g_2-1) = (1-g_1)(1-g_2).

Specific cases:

  • C1=C2=P1C_1 = C_2 = \mathbb{P}^1 (g=0g = 0): K2=8K^2 = 8, e=4e = 4, Ο‡=1\chi = 1. This is P1Γ—P1\mathbb{P}^1 \times \mathbb{P}^1.
  • C1=EC_1 = E, C2=Eβ€²C_2 = E' (elliptic, g=1g = 1): K2=0K^2 = 0, e=0e = 0, Ο‡=0\chi = 0. This is an abelian surface.
  • C1=C2C_1 = C_2 genus 22: K2=8K^2 = 8, e=4e = 4, q=4q = 4, pg=4p_g = 4, Ο‡=1\chi = 1.
  • C1=C2C_1 = C_2 genus 33: K2=32K^2 = 32, e=16e = 16, q=6q = 6, pg=9p_g = 9, Ο‡=4\chi = 4.
ExampleComplete intersections in projective space

Let X=Xd1,d2βŠ‚P4X = X_{d_1, d_2} \subset \mathbb{P}^4 be a smooth complete intersection of hypersurfaces of degrees d1,d2d_1, d_2. By adjunction:

KX=(d1+d2βˆ’5)H∣XK_X = (d_1 + d_2 - 5)H|_X, deg⁑(X)=d1d2\deg(X) = d_1 d_2, K2=d1d2(d1+d2βˆ’5)2K^2 = d_1 d_2(d_1 + d_2 - 5)^2.

The Euler characteristic is e(X)=d1d2(d12+d22+3d1d2βˆ’5d1βˆ’5d2+10)/1e(X) = d_1 d_2(d_1^2 + d_2^2 + 3d_1 d_2 - 5d_1 - 5d_2 + 10) / 1... more precisely, from the Chern class computation:

e(X)=d1d2(d12βˆ’5d1+d22βˆ’5d2+3d1d2+10)e(X) = d_1 d_2(d_1^2 - 5d_1 + d_2^2 - 5d_2 + 3d_1 d_2 + 10).

Specific cases:

  • (2,2)(2,2) in P4\mathbb{P}^4: K=βˆ’HK = -H, K2=4K^2 = 4, e=4β‹…(4βˆ’10+4βˆ’10+12+10)=4β‹…10=40e = 4 \cdot (4 - 10 + 4 - 10 + 12 + 10) = 4 \cdot 10 = 40... let me compute more carefully. The degree is 44. e=c2(TX)=4(4+4+12βˆ’10βˆ’10+10)=4β‹…10=40e = c_2(T_X) = 4(4+4+12-10-10+10) = 4 \cdot 10 = 40. But Ο‡=112(4+40)=4412\chi = \frac{1}{12}(4 + 40) = \frac{44}{12} which is not an integer. Let me recalculate.

For (2,2)(2,2) in P4\mathbb{P}^4: this is a del Pezzo surface of degree 44. It is the blowup of P2\mathbb{P}^2 at 55 points. So K2=9βˆ’5=4K^2 = 9 - 5 = 4, q=0q = 0, pg=0p_g = 0, Ο‡=1\chi = 1, e=12βˆ’4=8e = 12 - 4 = 8.

  • (2,3)(2,3) in P4\mathbb{P}^4: degree 66, K=0K = 0. This is a K3 surface. K2=0K^2 = 0, Ο‡=2\chi = 2, e=24e = 24.

  • (3,3)(3,3) in P4\mathbb{P}^4: degree 99, K=H∣XK = H|_X, K2=9K^2 = 9. Ο‡=112(9+e)\chi = \frac{1}{12}(9 + e). By explicit computation, q=0q = 0 and pg=4p_g = 4, so Ο‡=5\chi = 5. Then e=60βˆ’9=51e = 60 - 9 = 51.

  • (2,4)(2,4) in P4\mathbb{P}^4: degree 88, K=H∣XK = H|_X, K2=8K^2 = 8. We have q=0q = 0, pg=3p_g = 3, Ο‡=4\chi = 4, e=48βˆ’8=40e = 48 - 8 = 40.


Noether's inequality

Theorem5.9Noether's inequality

Let XX be a minimal surface of general type. Then:

KX2β‰₯2pgβˆ’4.K_X^2 \geq 2p_g - 4.

Equivalently, since 12Ο‡=K2+e12\chi = K^2 + e and eβ‰₯0e \geq 0 is not always true (but e>0e > 0 for surfaces of general type by Bogomolov--Miyaoka--Yau), the inequality bounds K2K^2 from below in terms of the geometric genus.

For irregular surfaces (q>0q > 0), the stronger bound K2β‰₯2Ο‡(OX)K^2 \geq 2\chi(\mathcal{O}_X) holds (the Bogomolov--Miyaoka--Yau inequality gives even more).

RemarkProof sketch of Noether's inequality

Consider the canonical map Ο•K ⁣:Xβ‡’Ppgβˆ’1\phi_K \colon X \dashrightarrow \mathbb{P}^{p_g - 1}.

Case 1: Ο•K\phi_K is birational onto its image Ξ£\Sigma. Then deg⁑(Ξ£)β‰₯pgβˆ’1\deg(\Sigma) \geq p_g - 1 (a nondegenerate variety in Ppgβˆ’1\mathbb{P}^{p_g-1} has degree at least pgβˆ’1p_g - 1). Since K2β‰₯deg⁑(Ξ£)β‰₯pgβˆ’1K^2 \geq \deg(\Sigma) \geq p_g - 1, we get K2β‰₯pgβˆ’1β‰₯2pgβˆ’4K^2 \geq p_g - 1 \geq 2p_g - 4 when pg≀3p_g \leq 3. For pgβ‰₯4p_g \geq 4, a more careful argument using the degree of the image as a surface in Ppgβˆ’1\mathbb{P}^{p_g - 1} gives K2β‰₯2pgβˆ’4K^2 \geq 2p_g - 4.

Case 2: Ο•K\phi_K has degree dβ‰₯2d \geq 2. Then K2=dβ‹…deg⁑(Ξ£)β‰₯2(pgβˆ’1)K^2 = d \cdot \deg(\Sigma) \geq 2(p_g - 1) if Ξ£\Sigma is a surface, giving K2β‰₯2pgβˆ’2>2pgβˆ’4K^2 \geq 2p_g - 2 > 2p_g - 4.

If ∣K∣|K| maps to a curve (the canonical system is composed with a pencil), a separate argument using fibrations gives the result.

ExampleChecking Noether's inequality
  • Quintic in P3\mathbb{P}^3 (d=5d = 5): K2=5K^2 = 5, pg=4p_g = 4. Inequality: 5β‰₯2(4)βˆ’4=45 \geq 2(4) - 4 = 4.
  • Product of two genus-22 curves: K2=8K^2 = 8, pg=4p_g = 4. Inequality: 8β‰₯48 \geq 4.
  • Godeaux surface (Ο€1=Z/5\pi_1 = \mathbb{Z}/5): K2=1K^2 = 1, pg=0p_g = 0. Inequality: 1β‰₯βˆ’41 \geq -4.
  • Barlow surface: K2=1K^2 = 1, pg=0p_g = 0, q=0q = 0. The simplest simply connected surface of general type.
  • Horikawa surfaces: these lie on the Noether line K2=2pgβˆ’4K^2 = 2p_g - 4, achieving equality. They exist for all pgβ‰₯3p_g \geq 3.

Bogomolov--Miyaoka--Yau inequality

Theorem5.10Bogomolov--Miyaoka--Yau inequality

Let XX be a minimal surface of general type. Then:

c12≀3c2,i.e.,KX2≀3e(X).c_1^2 \leq 3c_2, \quad \text{i.e.,} \quad K_X^2 \leq 3e(X).

Using Noether's formula 12Ο‡=c12+c212\chi = c_1^2 + c_2 to eliminate c2=12Ο‡βˆ’c12c_2 = 12\chi - c_1^2, this becomes:

c12≀3(12Ο‡βˆ’c12)=36Ο‡βˆ’3c12,hencec12≀9Ο‡(OX).c_1^2 \leq 3(12\chi - c_1^2) = 36\chi - 3c_1^2, \quad \text{hence} \quad c_1^2 \leq 9\chi(\mathcal{O}_X).

Equivalently, K2≀9Ο‡(OX)K^2 \leq 9\chi(\mathcal{O}_X).

Equality K2=9Ο‡K^2 = 9\chi holds if and only if the universal cover of XX is the complex unit ball B2\mathbb{B}^2 (so XX is a ball quotient).

RemarkOn the proof

The inequality c12≀3c2c_1^2 \leq 3c_2 was proved by Miyaoka (1977) using algebraic methods (Bogomolov--Miyaoka inequality for semistable bundles) and independently by Yau (1977) using differential geometry (the existence of Kahler--Einstein metrics on manifolds with ample canonical bundle, via the Calabi conjecture).

Yau's method shows that equality c12=3c2c_1^2 = 3c_2 implies the existence of a Kahler--Einstein metric with constant holomorphic sectional curvature, so the universal cover is B2\mathbb{B}^2.

ExampleChecking the BMY inequality
  • P2\mathbb{P}^2: c12=9c_1^2 = 9, c2=3c_2 = 3, ratio c12/c2=3c_1^2/c_2 = 3. This achieves equality! Indeed, P2=B2/Ξ“\mathbb{P}^2 = \mathbb{B}^2/\Gamma... no, P2\mathbb{P}^2 is simply connected and rational, not of general type. The BMY inequality applies only to surfaces of general type.
  • Quintic surface (d=5d = 5): K2=5K^2 = 5, e=55e = 55, Ο‡=5\chi = 5. Check: 5≀9β‹…5=455 \leq 9 \cdot 5 = 45 and 5≀3β‹…55=1655 \leq 3 \cdot 55 = 165.
  • Ball quotients: the fake projective planes have c12=3c2=9c_1^2 = 3c_2 = 9, Ο‡=1\chi = 1, pg=0p_g = 0, q=0q = 0. There are exactly 100100 fake projective planes (50 conjugate pairs), classified by Prasad--Yeung and Cartwright--Steger.
  • Product of two genus-22 curves: K2=8K^2 = 8, e=4e = 4, Ο‡=1\chi = 1. Check: 8≀3β‹…4=128 \leq 3 \cdot 4 = 12 and 8≀9β‹…1=98 \leq 9 \cdot 1 = 9.

Geography of surfaces

Definition5.4Geography problem

The geography of surfaces asks: which pairs (c12,c2)(c_1^2, c_2) (or equivalently (K2,Ο‡)(K^2, \chi)) are realized by minimal smooth projective surfaces of general type?

By Noether's formula, these pairs lie on the line c12+c2=12Ο‡c_1^2 + c_2 = 12\chi in the (c12,c2)(c_1^2, c_2)-plane. The geography is therefore determined by which points (K2,Ο‡)(K^2, \chi) are realized.

Theorem5.11Geography bounds

For a minimal surface of general type, the invariants (K2,Ο‡)(K^2, \chi) satisfy:

  • Noether line (lower bound): K2β‰₯2Ο‡βˆ’6K^2 \geq 2\chi - 6 (equivalently c12β‰₯2c2/5c_1^2 \geq 2c_2/5, or from K2β‰₯2pgβˆ’4K^2 \geq 2p_g - 4).
  • BMY line (upper bound): K2≀9Ο‡K^2 \leq 9\chi (equivalently c12≀3c2c_1^2 \leq 3c_2).
  • Positivity: K2β‰₯1K^2 \geq 1 and Ο‡β‰₯1\chi \geq 1.
  • Integrality: K2K^2 and Ο‡\chi are positive integers with K2+e=12Ο‡K^2 + e = 12\chi where e>0e > 0.

The "geography region" is the set of lattice points (K2,Ο‡)(K^2, \chi) with 2Ο‡βˆ’6≀K2≀9Ο‡2\chi - 6 \leq K^2 \leq 9\chi and K2,Ο‡β‰₯1K^2, \chi \geq 1.

ExampleKnown surfaces in the geography

Key surfaces and their positions in the (K2,Ο‡)(K^2, \chi)-plane:

On or near the Noether line (K2=2Ο‡βˆ’6K^2 = 2\chi - 6):

  • Horikawa surfaces: K2=2pgβˆ’4K^2 = 2p_g - 4, ranging over K2=2,4,6,…K^2 = 2, 4, 6, \ldots with Ο‡=(K2+6)/2=4,5,6,…\chi = (K^2 + 6)/2 = 4, 5, 6, \ldots.

On the BMY line (K2=9Ο‡K^2 = 9\chi):

  • Fake projective planes: (K2,Ο‡)=(9,1)(K^2, \chi) = (9, 1).
  • Ball quotients with Ο‡=2\chi = 2: (K2,Ο‡)=(18,2)(K^2, \chi) = (18, 2).

Between the lines:

  • Godeaux surface: (K2,Ο‡)=(1,1)(K^2, \chi) = (1, 1).
  • Campedelli surface: (K2,Ο‡)=(2,1)(K^2, \chi) = (2, 1).
  • Barlow surface: (K2,Ο‡)=(1,1)(K^2, \chi) = (1, 1) (same invariants as Godeaux, but simply connected).
  • Catanese surface: (K2,Ο‡)=(2,1)(K^2, \chi) = (2, 1).
  • Burniat surfaces: (K2,Ο‡)=(2,1),(3,1),(4,1),(5,1),(6,1)(K^2, \chi) = (2, 1), (3, 1), (4, 1), (5, 1), (6, 1).
  • Beauville surface: (K2,Ο‡)=(8,1)(K^2, \chi) = (8, 1).
  • Quintic in P3\mathbb{P}^3: (K2,Ο‡)=(5,5)(K^2, \chi) = (5, 5).
RemarkDensity of the geography

Persson (1981) showed that almost all lattice points in the geography region are realized. More precisely: for every Ο΅>0\epsilon > 0, the lattice points (K2,Ο‡)(K^2, \chi) with 2Ο‡βˆ’6≀K2≀(9βˆ’Ο΅)Ο‡2\chi - 6 \leq K^2 \leq (9 - \epsilon)\chi are realized for χ≫0\chi \gg 0.

Near the BMY line, the geography is sparse: very few surfaces have K2/Ο‡K^2/\chi close to 99. Ball quotients are "rare" arithmetic objects.

Near the Noether line, Horikawa surfaces fill out a dense family.


Chern numbers by Kodaira dimension

Definition5.5Kodaira dimension

The Kodaira dimension κ(X)\kappa(X) is the growth rate of the pluricanonical system ∣nKX∣|nK_X|:

  • ΞΊ=βˆ’βˆž\kappa = -\infty: Pn=h0(nKX)=0P_n = h^0(nK_X) = 0 for all nβ‰₯1n \geq 1.
  • ΞΊ=0\kappa = 0: PnP_n is bounded but not identically 00.
  • ΞΊ=1\kappa = 1: PnP_n grows linearly.
  • ΞΊ=2\kappa = 2: PnP_n grows quadratically (surfaces of general type).
ExampleKodaira dimension -infinity (rational and ruled)

Rational surfaces (XX birational to P2\mathbb{P}^2):

  • Ο‡(OX)=1\chi(\mathcal{O}_X) = 1, q=0q = 0, pg=0p_g = 0.
  • Minimal models: P2\mathbb{P}^2 (K2=9,e=3K^2 = 9, e = 3) and Fn\mathbb{F}_n for nβ‰ 1n \neq 1 (K2=8,e=4K^2 = 8, e = 4).
  • Blowups: Bl⁑r(P2)\operatorname{Bl}_r(\mathbb{P}^2) has K2=9βˆ’rK^2 = 9 - r, e=3+re = 3 + r.

Ruled surfaces over a curve of genus gg (geometrically ruled: X=P(E)β†’CX = \mathbb{P}(\mathcal{E}) \to C):

  • q=gq = g, pg=0p_g = 0, Ο‡(OX)=1βˆ’g\chi(\mathcal{O}_X) = 1 - g.
  • K2=8(1βˆ’g)K^2 = 8(1 - g), e=4(1βˆ’g)e = 4(1 - g).
  • Noether: 12(1βˆ’g)=8(1βˆ’g)+4(1βˆ’g)12(1 - g) = 8(1 - g) + 4(1 - g).
ExampleKodaira dimension 0

The minimal surfaces with ΞΊ=0\kappa = 0 fall into four classes:

K3 surfaces: q=0q = 0, pg=1p_g = 1, Ο‡=2\chi = 2, K2=0K^2 = 0, e=24e = 24.

Enriques surfaces: q=0q = 0, pg=0p_g = 0, Ο‡=1\chi = 1, K2=0K^2 = 0, e=12e = 12. (In characteristic β‰ 2\neq 2, 2K∼02K \sim 0.)

Abelian surfaces: q=2q = 2, pg=1p_g = 1, Ο‡=0\chi = 0, K2=0K^2 = 0, e=0e = 0.

Hyperelliptic (bielliptic) surfaces: q=1q = 1, pg=0p_g = 0, Ο‡=0\chi = 0, K2=0K^2 = 0, e=0e = 0. (These are quotients (E1Γ—E2)/G(E_1 \times E_2)/G where GG is a finite group acting on both elliptic curves.)

In all cases K2=0K^2 = 0. Noether gives Ο‡=e/12\chi = e/12, so ee is a multiple of 1212 for Ο‡βˆˆZ\chi \in \mathbb{Z}. Indeed: e∈{0,0,12,24}e \in \{0, 0, 12, 24\}.

ExampleKodaira dimension 1 (elliptic surfaces)

Minimal surfaces with ΞΊ=1\kappa = 1 are properly elliptic surfaces: they admit a fibration f ⁣:Xβ†’Cf \colon X \to C whose general fiber is an elliptic curve, and KXK_X is not numerically trivial.

For a relatively minimal elliptic fibration f ⁣:Xβ†’Cf \colon X \to C with g(C)=gg(C) = g:

  • Ο‡(OX)=Ο‡(OC)β‹…Ο‡(OF)+(singularΒ fiberΒ contribution)=deg⁑fβˆ—Ο‰X/C\chi(\mathcal{O}_X) = \chi(\mathcal{O}_C) \cdot \chi(\mathcal{O}_F) + \text{(singular fiber contribution)} = \deg f_* \omega_{X/C}.
  • K2=0K^2 = 0 (for minimal models).
  • e(X)=12Ο‡(OX)e(X) = 12\chi(\mathcal{O}_X) (by Noether, since K2=0K^2 = 0).

The Euler characteristic e(X)e(X) is computed from the singular fibers via Ogg's formula: e(X)=βˆ‘ve(Xv)e(X) = \sum_v e(X_v) where the sum is over singular fibers.

Example: Elliptic surface over P1\mathbb{P}^1 with Ο‡=1\chi = 1: K2=0K^2 = 0, e=12e = 12. If it has 1212 nodal fibers (type I1I_1), each contributes e=1e = 1 to the Euler characteristic: e=12β‹…1=12e = 12 \cdot 1 = 12.

Example: Elliptic K3 surface: Ο‡=2\chi = 2, e=24e = 24. Possible singular fiber configuration: 2424 nodal fibers (all I1I_1), or two fibers of type I0βˆ—I_0^* (each contributes e=6e = 6, giving 1212) plus 1212 nodal fibers, etc.


The signature and Hirzebruch's theorem

Definition5.6Signature of a surface

For a smooth compact complex surface XX, the signature (or index) Οƒ(X)\sigma(X) is the signature of the intersection form on H2(X,R)H^2(X, \mathbb{R}):

Οƒ(X)=b2+βˆ’b2βˆ’,\sigma(X) = b_2^+ - b_2^-,

where b2+b_2^+ (resp. b2βˆ’b_2^-) is the number of positive (resp. negative) eigenvalues of the intersection form.

Theorem5.12Hirzebruch signature theorem for surfaces

For a smooth compact complex surface XX:

Οƒ(X)=13(c12βˆ’2c2)=13(KX2βˆ’2e(X)).\sigma(X) = \frac{1}{3}(c_1^2 - 2c_2) = \frac{1}{3}(K_X^2 - 2e(X)).

Combined with Noether's formula 12Ο‡=c12+c212\chi = c_1^2 + c_2, one can express c12c_1^2 and c2c_2 separately:

c12=K2=12Ο‡βˆ’e,c2=e,Οƒ=13(12Ο‡βˆ’3e)=4Ο‡βˆ’e.c_1^2 = K^2 = 12\chi - e, \quad c_2 = e, \quad \sigma = \frac{1}{3}(12\chi - 3e) = 4\chi - e.

Equivalently: Οƒ(X)=4Ο‡(OX)βˆ’e(X)\sigma(X) = 4\chi(\mathcal{O}_X) - e(X).

In terms of Hodge numbers: Οƒ(X)=(2h2,0+1)βˆ’h1,1=2pg+1βˆ’h1,1=4Ο‡βˆ’e\sigma(X) = (2h^{2,0} + 1) - h^{1,1} = 2p_g + 1 - h^{1,1} = 4\chi - e (using b2+=2pg+1b_2^+ = 2p_g + 1, b2βˆ’=h1,1βˆ’1b_2^- = h^{1,1} - 1 by the Hodge index theorem).

ExampleComputing signatures
  • P2\mathbb{P}^2: Οƒ=4(1)βˆ’3=1\sigma = 4(1) - 3 = 1. Intersection form on H2H^2: the 1Γ—11 \times 1 matrix (1)(1), signature +1+1.
  • P1Γ—P1\mathbb{P}^1 \times \mathbb{P}^1: Οƒ=4(1)βˆ’4=0\sigma = 4(1) - 4 = 0. Intersection form: the hyperbolic form HH with eigenvalues Β±1\pm 1, signature 00.
  • K3 surface: Οƒ=4(2)βˆ’24=βˆ’16\sigma = 4(2) - 24 = -16. The intersection form is 3HβŠ•(βˆ’E8)23H \oplus (-E_8)^2 (three hyperbolic planes plus two copies of the negative E8E_8 lattice), with b2+=3,b2βˆ’=19b_2^+ = 3, b_2^- = 19, signature 3βˆ’19=βˆ’163 - 19 = -16.
  • Abelian surface: Οƒ=4(0)βˆ’0=0\sigma = 4(0) - 0 = 0. The intersection form on H2(Z)β‰…Z6H^2(\mathbb{Z}) \cong \mathbb{Z}^6 has b2+=3,b2βˆ’=3b_2^+ = 3, b_2^- = 3, signature 00.
  • Enriques surface: Οƒ=4(1)βˆ’12=βˆ’8\sigma = 4(1) - 12 = -8. The intersection form is HβŠ•(βˆ’E8)H \oplus (-E_8) (one hyperbolic plane plus the negative E8E_8 lattice), with b2+=1,b2βˆ’=9b_2^+ = 1, b_2^- = 9, signature βˆ’8-8.
  • Quintic surface: Οƒ=4(5)βˆ’55=βˆ’35\sigma = 4(5) - 55 = -35.

Refined invariants: the Chern number table

ExampleComplete Chern number table

A reference table of invariants for the most important surfaces.

Rational surfaces:

  • P2\mathbb{P}^2: K2=9K^2 = 9, e=3e = 3, Ο‡=1\chi = 1, q=0q = 0, pg=0p_g = 0, Οƒ=1\sigma = 1, b2=1b_2 = 1.
  • P1Γ—P1\mathbb{P}^1 \times \mathbb{P}^1: K2=8K^2 = 8, e=4e = 4, Ο‡=1\chi = 1, q=0q = 0, pg=0p_g = 0, Οƒ=0\sigma = 0, b2=2b_2 = 2.
  • Bl⁑1(P2)\operatorname{Bl}_1(\mathbb{P}^2): K2=8K^2 = 8, e=4e = 4, Ο‡=1\chi = 1, q=0q = 0, pg=0p_g = 0, Οƒ=0\sigma = 0, b2=2b_2 = 2.
  • Bl⁑6(P2)\operatorname{Bl}_6(\mathbb{P}^2) (cubic surface): K2=3K^2 = 3, e=9e = 9, Ο‡=1\chi = 1, q=0q = 0, pg=0p_g = 0, Οƒ=βˆ’5\sigma = -5, b2=7b_2 = 7.
  • Bl⁑8(P2)\operatorname{Bl}_8(\mathbb{P}^2): K2=1K^2 = 1, e=11e = 11, Ο‡=1\chi = 1, Οƒ=βˆ’7\sigma = -7, b2=9b_2 = 9.

Kodaira dimension 0:

  • K3: K2=0K^2 = 0, e=24e = 24, Ο‡=2\chi = 2, q=0q = 0, pg=1p_g = 1, Οƒ=βˆ’16\sigma = -16, b2=22b_2 = 22.
  • Enriques: K2=0K^2 = 0, e=12e = 12, Ο‡=1\chi = 1, q=0q = 0, pg=0p_g = 0, Οƒ=βˆ’8\sigma = -8, b2=10b_2 = 10.
  • Abelian: K2=0K^2 = 0, e=0e = 0, Ο‡=0\chi = 0, q=2q = 2, pg=1p_g = 1, Οƒ=0\sigma = 0, b2=6b_2 = 6.
  • Hyperelliptic: K2=0K^2 = 0, e=0e = 0, Ο‡=0\chi = 0, q=1q = 1, pg=0p_g = 0, Οƒ=0\sigma = 0, b2=2b_2 = 2.

Surfaces of general type:

  • Godeaux (K2=1K^2 = 1): e=11e = 11, Ο‡=1\chi = 1, Οƒ=βˆ’7\sigma = -7.
  • Campedelli (K2=2K^2 = 2): e=10e = 10, Ο‡=1\chi = 1, Οƒ=βˆ’6\sigma = -6.
  • Quintic (K2=5K^2 = 5): e=55e = 55, Ο‡=5\chi = 5, Οƒ=βˆ’35\sigma = -35.
  • Fake P2\mathbb{P}^2 (K2=9K^2 = 9): e=3e = 3, Ο‡=1\chi = 1, Οƒ=1\sigma = 1.

Constraints from Noether's formula

RemarkArithmetic consequences

Noether's formula 12Ο‡=K2+e12\chi = K^2 + e imposes strong arithmetic constraints:

  1. Integrality of Ο‡\chi: Since K2K^2 and ee are integers, K2+e≑0(mod12)K^2 + e \equiv 0 \pmod{12}.

  2. Parity: Over C\mathbb{C}, Wu's formula gives Kβ‹…D≑D2(mod2)K \cdot D \equiv D^2 \pmod{2} for all DD. In particular K2≑Kβ‹…K≑K2(mod2)K^2 \equiv K \cdot K \equiv K^2 \pmod{2} (tautological), but also K2+e≑0(mod12)K^2 + e \equiv 0 \pmod{12} constrains parity relations between K2K^2 and ee.

  3. For spin surfaces (those with K≑0(mod2)K \equiv 0 \pmod{2} in H2(X,Z/2)H^2(X, \mathbb{Z}/2)): Rokhlin's theorem gives Οƒ(X)≑0(mod16)\sigma(X) \equiv 0 \pmod{16}, i.e., 4Ο‡βˆ’e≑0(mod16)4\chi - e \equiv 0 \pmod{16}. Combined with 12Ο‡=K2+e12\chi = K^2 + e, this constrains the invariants further.

ExampleSpin condition for K3 and complete intersections

K3 surfaces are spin (since K=0≑0(mod2)K = 0 \equiv 0 \pmod{2}). Rokhlin: Οƒ=βˆ’16≑0(mod16)\sigma = -16 \equiv 0 \pmod{16}.

Complete intersection (2,3)(2,3) in P4\mathbb{P}^4 (K3): same as above.

Quartic K3 in P3\mathbb{P}^3: K=0K = 0, spin, Οƒ=βˆ’16\sigma = -16.

Quintic surface in P3\mathbb{P}^3: K=HK = H, and HH generates H2(X,Z/2)H^2(X, \mathbb{Z}/2), so this is not spin. Οƒ=βˆ’35\sigma = -35, which is not divisible by 1616.

Sextic surface (d=6d = 6): K=2HK = 2H, so K≑0(mod2)K \equiv 0 \pmod{2}, this is spin. K2=24K^2 = 24, Ο‡=11\chi = 11, e=108e = 108, Οƒ=4(11)βˆ’108=βˆ’64=βˆ’4β‹…16\sigma = 4(11) - 108 = -64 = -4 \cdot 16. Rokhlin's congruence is satisfied.


Relation to the four-manifold invariants

RemarkNoether's formula in topology

For a simply connected smooth compact complex surface XX, Freedman's theorem says the homeomorphism type is determined by the intersection form on H2(X,Z)H^2(X, \mathbb{Z}).

Noether's formula constrains this form: the rank is b2=eβˆ’2b_2 = e - 2, and the signature is Οƒ=4Ο‡βˆ’e=(K2βˆ’2e)/3\sigma = 4\chi - e = (K^2 - 2e)/3.

For a minimal surface of general type:

  • b2+=2pg+1b_2^+ = 2p_g + 1 (odd, since pgβ‰₯0p_g \geq 0).
  • b2βˆ’=b2βˆ’b2+=(eβˆ’2)βˆ’(2pg+1)=eβˆ’2pgβˆ’3b_2^- = b_2 - b_2^+ = (e - 2) - (2p_g + 1) = e - 2p_g - 3.
  • Noether: b2=eβˆ’2=12Ο‡βˆ’K2βˆ’2b_2 = e - 2 = 12\chi - K^2 - 2.

The Donaldson and Seiberg--Witten invariants provide smooth-structure information beyond the topological data. Surfaces of general type are "standard" from the Seiberg--Witten perspective: they have simple type and the basic classes are Β±KX\pm K_X.


Further examples

ExampleDouble covers of P2

Let X→P2X \to \mathbb{P}^2 be a double cover branched along a smooth curve BB of degree 2d2d. Then:

KX=Ο€βˆ—(KP2+dH)=Ο€βˆ—((dβˆ’3)H)K_X = \pi^*(K_{\mathbb{P}^2} + dH) = \pi^*((d - 3)H), so K2=2(dβˆ’3)2K^2 = 2(d-3)^2.

e(X)=2e(P2)βˆ’e(B)=2β‹…3βˆ’(2βˆ’2g(B))=6βˆ’2+2β‹…(2dβˆ’1)(2dβˆ’2)2=4+2d(2dβˆ’3)e(X) = 2e(\mathbb{P}^2) - e(B) = 2 \cdot 3 - (2 - 2g(B)) = 6 - 2 + 2 \cdot \frac{(2d-1)(2d-2)}{2} = 4 + 2d(2d - 3)... let me compute more carefully. The branch curve BB has degree 2d2d and genus g(B)=(2dβˆ’1)(2dβˆ’2)2=(2dβˆ’1)(dβˆ’1)g(B) = \frac{(2d-1)(2d-2)}{2} = (2d-1)(d-1). The Euler characteristic of the double cover is:

e(X)=2e(P2βˆ–B)+e(B)=2(3βˆ’e(B))+e(B)=6βˆ’e(B)=6βˆ’(2βˆ’2g(B))=4+2(2dβˆ’1)(dβˆ’1)e(X) = 2e(\mathbb{P}^2 \setminus B) + e(B) = 2(3 - e(B)) + e(B) = 6 - e(B) = 6 - (2 - 2g(B)) = 4 + 2(2d-1)(d-1).

Simplifying: e(X)=4d2βˆ’6d+6e(X) = 4d^2 - 6d + 6.

Noether check: 112(2(dβˆ’3)2+4d2βˆ’6d+6)=112(2d2βˆ’12d+18+4d2βˆ’6d+6)=112(6d2βˆ’18d+24)=d2βˆ’3d+42\frac{1}{12}(2(d-3)^2 + 4d^2 - 6d + 6) = \frac{1}{12}(2d^2 - 12d + 18 + 4d^2 - 6d + 6) = \frac{1}{12}(6d^2 - 18d + 24) = \frac{d^2 - 3d + 4}{2}.

And indeed Ο‡(OX)=1βˆ’0+pg\chi(\mathcal{O}_X) = 1 - 0 + p_g where pg=h0(KX)=h0(OP2(dβˆ’3))+h0(OP2(βˆ’3))=(dβˆ’12)p_g = h^0(K_X) = h^0(\mathcal{O}_{\mathbb{P}^2}(d-3)) + h^0(\mathcal{O}_{\mathbb{P}^2}(-3)) = \binom{d-1}{2} for dβ‰₯3d \geq 3. With q=0q = 0:

Ο‡=1+(dβˆ’12)=1+(dβˆ’1)(dβˆ’2)2=d2βˆ’3d+42\chi = 1 + \binom{d-1}{2} = 1 + \frac{(d-1)(d-2)}{2} = \frac{d^2 - 3d + 4}{2}.

Specific cases:

  • d=3d = 3 (BB sextic): K2=0K^2 = 0, e=24e = 24, Ο‡=2\chi = 2. This is a K3 surface.
  • d=4d = 4 (BB octic): K2=2K^2 = 2, e=46e = 46, Ο‡=4\chi = 4, pg=3p_g = 3.
  • d=5d = 5 (BB degree 1010): K2=8K^2 = 8, e=76e = 76, Ο‡=7\chi = 7, pg=6p_g = 6.
ExampleFibered surfaces

Let f ⁣:Xβ†’Cf \colon X \to C be a fibration with general fiber FF of genus g(F)=gg(F) = g over a base curve of genus bb.

The Chern numbers satisfy:

  • e(X)=4(gβˆ’1)(bβˆ’1)+βˆ‘vΞ΄ve(X) = 4(g-1)(b-1) + \sum_v \delta_v where Ξ΄vβ‰₯0\delta_v \geq 0 measures the singular fibers.
  • K2=8(gβˆ’1)(bβˆ’1)+KX/C2K^2 = 8(g-1)(b-1) + K_{X/C}^2 where KX/C2β‰₯0K_{X/C}^2 \geq 0 by the Arakelov inequality (and equals 00 iff the fibration is isotrivial after base change).
  • Ο‡(OX)=(gβˆ’1)(bβˆ’1)+deg⁑fβˆ—Ο‰X/C\chi(\mathcal{O}_X) = (g - 1)(b - 1) + \deg f_* \omega_{X/C}.

For a Kodaira fibration (a smooth fibration with no singular fibers, but non-isotrivial):

  • K2>8(gβˆ’1)(bβˆ’1)>0K^2 > 8(g-1)(b-1) > 0 (strict inequality by the Arakelov inequality).
  • e=4(gβˆ’1)(bβˆ’1)e = 4(g-1)(b-1).
  • These satisfy K2/e>2K^2/e > 2, giving points in the geography above the line c12=2c2c_1^2 = 2c_2.

Universality of Noether's formula

RemarkNoether's formula in higher dimensions

Noether's formula generalizes to higher dimensions via the Hirzebruch--Riemann--Roch theorem:

Ο‡(OX)=∫Xtd⁑(TX).\chi(\mathcal{O}_X) = \int_X \operatorname{td}(T_X).

In dimension 11 (curves): Ο‡(OC)=1βˆ’g=12(c1(TC))+1=12(2βˆ’2g)+0\chi(\mathcal{O}_C) = 1 - g = \frac{1}{2}(c_1(T_C)) + 1 = \frac{1}{2}(2 - 2g) + 0... more precisely, Ο‡(OC)=12deg⁑(KC)+1=1βˆ’g\chi(\mathcal{O}_C) = \frac{1}{2}\deg(K_C) + 1 = 1 - g.

In dimension 22 (surfaces): Ο‡(OX)=112(c12+c2)\chi(\mathcal{O}_X) = \frac{1}{12}(c_1^2 + c_2) -- Noether's formula.

In dimension 33 (threefolds): Ο‡(OX)=124c1c2\chi(\mathcal{O}_X) = \frac{1}{24}c_1 c_2.

In dimension 44 (fourfolds): Ο‡(OX)=1720(3c14βˆ’c12c2βˆ’3c22+c4)\chi(\mathcal{O}_X) = \frac{1}{720}(3c_1^4 - c_1^2 c_2 - 3c_2^2 + c_4)... more precisely, Ο‡(OX)=1720(βˆ’c14+4c12c2+3c22+c1c3βˆ’c4)\chi(\mathcal{O}_X) = \frac{1}{720}(-c_1^4 + 4c_1^2 c_2 + 3c_2^2 + c_1 c_3 - c_4).

The surface case is distinguished by its simplicity: only two Chern numbers (c12c_1^2 and c2c_2) constrained by one relation (12Ο‡=c12+c212\chi = c_1^2 + c_2), leaving a one-dimensional "geography."


Summary

RemarkThe central role of Noether's formula

Noether's formula 12Ο‡(OX)=KX2+e(X)12\chi(\mathcal{O}_X) = K_X^2 + e(X) is the single most important numerical identity in surface theory. It:

  • Constrains invariants: any two of {K2,e,Ο‡}\{K^2, e, \chi\} determine the third.
  • Underlies Riemann--Roch: the Riemann--Roch theorem for surfaces Ο‡(O(D))=12D(Dβˆ’K)+Ο‡(OX)\chi(\mathcal{O}(D)) = \frac{1}{2}D(D - K) + \chi(\mathcal{O}_X) uses Ο‡(OX)\chi(\mathcal{O}_X) as an input, and Noether's formula expresses it in terms of Chern numbers.
  • Bounds geography: combined with the Noether inequality (K2β‰₯2Ο‡βˆ’6K^2 \geq 2\chi - 6) and BMY inequality (K2≀9Ο‡K^2 \leq 9\chi), it delineates the region of possible invariants.
  • Classifies surfaces: the four classes of ΞΊ=0\kappa = 0 surfaces are distinguished by their Chern numbers.
  • Connects topology and algebra: it relates the holomorphic invariant Ο‡(OX)\chi(\mathcal{O}_X) to the topological invariants K2K^2 and ee.