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
Let be a smooth projective surface over an algebraically closed field. Then:
where is the holomorphic Euler characteristic, is the self-intersection of the canonical divisor, and is the topological Euler characteristic.
Equivalently, in terms of Chern numbers:
Here:
- is the irregularity.
- is the geometric genus.
- so (the first Chern class of the cotangent bundle equals the canonical class).
- (the second Chern class equals the topological Euler characteristic).
- Over , we have where are the Betti numbers.
Derivation from Hirzebruch--Riemann--Roch
For a smooth projective surface and a vector bundle (or coherent sheaf) :
The Todd class of the tangent bundle is .
Setting (the trivial line bundle, ):
, so .
The integral picks out the degree- component of the Todd class:
Since , we have , and . This gives:
Noether's formula is simply HRR applied to the structure sheaf.
Over the complex numbers
Over , the Hodge decomposition gives with . The Betti numbers are .
For a surface: , , , , .
The topological Euler characteristic is:
Substituting into Noether's formula and writing :
which simplifies to , or equivalently:
This determines from , , and .
Verification: classical examples
For : , so . The Betti numbers are , giving .
Noether: .
Direct computation: , , , so .
For : , so . The Betti numbers are , giving .
Noether: .
Direct: , , so .
A K3 surface has (so ), , and .
Noether: , hence .
Betti numbers: . Check: .
Hodge diamond: , , , .
From the formula: ... let us recheck: . But . So indeed .
An abelian surface has (so ), , .
Noether: , hence .
Betti numbers: . Check: .
Hodge diamond: , , , .
An Enriques surface has but , , .
(since implies ).
Noether: , hence .
Betti numbers: . Check: .
Let be a smooth hypersurface of degree . Then by adjunction, so .
The Euler characteristic of is computed from the Chern classes of restricted to . Using the exact sequence :
Noether: ... let us compute directly:
.
Specific values:
- (plane): , , .
- (quadric): , , .
- (cubic): , , .
- (quartic = K3): , , .
- (quintic): , , .
The cubic surface is rational: it is the blowup of at points. So , , .
, so (the degree). But for the cubic surface.
From the blowup description: , and .
. Alternatively: (the class of the line plus exceptional divisors), so .
Let where has genus . Then:
- , so .
- .
- , .
- .
Noether check: .
Specific cases:
- (): , , . This is .
- , (elliptic, ): , , . This is an abelian surface.
- genus : , , , , .
- genus : , , , , .
Let be a smooth complete intersection of hypersurfaces of degrees . By adjunction:
, , .
The Euler characteristic is ... more precisely, from the Chern class computation:
.
Specific cases:
- in : , , ... let me compute more carefully. The degree is . . But which is not an integer. Let me recalculate.
For in : this is a del Pezzo surface of degree . It is the blowup of at points. So , , , , .
-
in : degree , . This is a K3 surface. , , .
-
in : degree , , . . By explicit computation, and , so . Then .
-
in : degree , , . We have , , , .
Noether's inequality
Let be a minimal surface of general type. Then:
Equivalently, since and is not always true (but for surfaces of general type by Bogomolov--Miyaoka--Yau), the inequality bounds from below in terms of the geometric genus.
For irregular surfaces (), the stronger bound holds (the Bogomolov--Miyaoka--Yau inequality gives even more).
Consider the canonical map .
Case 1: is birational onto its image . Then (a nondegenerate variety in has degree at least ). Since , we get when . For , a more careful argument using the degree of the image as a surface in gives .
Case 2: has degree . Then if is a surface, giving .
If maps to a curve (the canonical system is composed with a pencil), a separate argument using fibrations gives the result.
- Quintic in (): , . Inequality: .
- Product of two genus- curves: , . Inequality: .
- Godeaux surface (): , . Inequality: .
- Barlow surface: , , . The simplest simply connected surface of general type.
- Horikawa surfaces: these lie on the Noether line , achieving equality. They exist for all .
Bogomolov--Miyaoka--Yau inequality
Let be a minimal surface of general type. Then:
Using Noether's formula to eliminate , this becomes:
Equivalently, .
Equality holds if and only if the universal cover of is the complex unit ball (so is a ball quotient).
The inequality 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 implies the existence of a Kahler--Einstein metric with constant holomorphic sectional curvature, so the universal cover is .
- : , , ratio . This achieves equality! Indeed, ... no, is simply connected and rational, not of general type. The BMY inequality applies only to surfaces of general type.
- Quintic surface (): , , . Check: and .
- Ball quotients: the fake projective planes have , , , . There are exactly fake projective planes (50 conjugate pairs), classified by Prasad--Yeung and Cartwright--Steger.
- Product of two genus- curves: , , . Check: and .
Geography of surfaces
The geography of surfaces asks: which pairs (or equivalently ) are realized by minimal smooth projective surfaces of general type?
By Noether's formula, these pairs lie on the line in the -plane. The geography is therefore determined by which points are realized.
For a minimal surface of general type, the invariants satisfy:
- Noether line (lower bound): (equivalently , or from ).
- BMY line (upper bound): (equivalently ).
- Positivity: and .
- Integrality: and are positive integers with where .
The "geography region" is the set of lattice points with and .
Key surfaces and their positions in the -plane:
On or near the Noether line ():
- Horikawa surfaces: , ranging over with .
On the BMY line ():
- Fake projective planes: .
- Ball quotients with : .
Between the lines:
- Godeaux surface: .
- Campedelli surface: .
- Barlow surface: (same invariants as Godeaux, but simply connected).
- Catanese surface: .
- Burniat surfaces: .
- Beauville surface: .
- Quintic in : .
Persson (1981) showed that almost all lattice points in the geography region are realized. More precisely: for every , the lattice points with are realized for .
Near the BMY line, the geography is sparse: very few surfaces have close to . Ball quotients are "rare" arithmetic objects.
Near the Noether line, Horikawa surfaces fill out a dense family.
Chern numbers by Kodaira dimension
The Kodaira dimension is the growth rate of the pluricanonical system :
- : for all .
- : is bounded but not identically .
- : grows linearly.
- : grows quadratically (surfaces of general type).
Rational surfaces ( birational to ):
- , , .
- Minimal models: () and for ().
- Blowups: has , .
Ruled surfaces over a curve of genus (geometrically ruled: ):
- , , .
- , .
- Noether: .
The minimal surfaces with fall into four classes:
K3 surfaces: , , , , .
Enriques surfaces: , , , , . (In characteristic , .)
Abelian surfaces: , , , , .
Hyperelliptic (bielliptic) surfaces: , , , , . (These are quotients where is a finite group acting on both elliptic curves.)
In all cases . Noether gives , so is a multiple of for . Indeed: .
Minimal surfaces with are properly elliptic surfaces: they admit a fibration whose general fiber is an elliptic curve, and is not numerically trivial.
For a relatively minimal elliptic fibration with :
- .
- (for minimal models).
- (by Noether, since ).
The Euler characteristic is computed from the singular fibers via Ogg's formula: where the sum is over singular fibers.
Example: Elliptic surface over with : , . If it has nodal fibers (type ), each contributes to the Euler characteristic: .
Example: Elliptic K3 surface: , . Possible singular fiber configuration: nodal fibers (all ), or two fibers of type (each contributes , giving ) plus nodal fibers, etc.
The signature and Hirzebruch's theorem
For a smooth compact complex surface , the signature (or index) is the signature of the intersection form on :
where (resp. ) is the number of positive (resp. negative) eigenvalues of the intersection form.
For a smooth compact complex surface :
Combined with Noether's formula , one can express and separately:
Equivalently: .
In terms of Hodge numbers: (using , by the Hodge index theorem).
- : . Intersection form on : the matrix , signature .
- : . Intersection form: the hyperbolic form with eigenvalues , signature .
- K3 surface: . The intersection form is (three hyperbolic planes plus two copies of the negative lattice), with , signature .
- Abelian surface: . The intersection form on has , signature .
- Enriques surface: . The intersection form is (one hyperbolic plane plus the negative lattice), with , signature .
- Quintic surface: .
Refined invariants: the Chern number table
A reference table of invariants for the most important surfaces.
Rational surfaces:
- : , , , , , , .
- : , , , , , , .
- : , , , , , , .
- (cubic surface): , , , , , , .
- : , , , , .
Kodaira dimension 0:
- K3: , , , , , , .
- Enriques: , , , , , , .
- Abelian: , , , , , , .
- Hyperelliptic: , , , , , , .
Surfaces of general type:
- Godeaux (): , , .
- Campedelli (): , , .
- Quintic (): , , .
- Fake (): , , .
Constraints from Noether's formula
Noether's formula imposes strong arithmetic constraints:
-
Integrality of : Since and are integers, .
-
Parity: Over , Wu's formula gives for all . In particular (tautological), but also constrains parity relations between and .
-
For spin surfaces (those with in ): Rokhlin's theorem gives , i.e., . Combined with , this constrains the invariants further.
K3 surfaces are spin (since ). Rokhlin: .
Complete intersection in (K3): same as above.
Quartic K3 in : , spin, .
Quintic surface in : , and generates , so this is not spin. , which is not divisible by .
Sextic surface (): , so , this is spin. , , , . Rokhlin's congruence is satisfied.
Relation to the four-manifold invariants
For a simply connected smooth compact complex surface , Freedman's theorem says the homeomorphism type is determined by the intersection form on .
Noether's formula constrains this form: the rank is , and the signature is .
For a minimal surface of general type:
- (odd, since ).
- .
- Noether: .
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 .
Further examples
Let be a double cover branched along a smooth curve of degree . Then:
, so .
... let me compute more carefully. The branch curve has degree and genus . The Euler characteristic of the double cover is:
.
Simplifying: .
Noether check: .
And indeed where for . With :
.
Specific cases:
- ( sextic): , , . This is a K3 surface.
- ( octic): , , , .
- ( degree ): , , , .
Let be a fibration with general fiber of genus over a base curve of genus .
The Chern numbers satisfy:
- where measures the singular fibers.
- where by the Arakelov inequality (and equals iff the fibration is isotrivial after base change).
- .
For a Kodaira fibration (a smooth fibration with no singular fibers, but non-isotrivial):
- (strict inequality by the Arakelov inequality).
- .
- These satisfy , giving points in the geography above the line .
Universality of Noether's formula
Noether's formula generalizes to higher dimensions via the Hirzebruch--Riemann--Roch theorem:
In dimension (curves): ... more precisely, .
In dimension (surfaces): -- Noether's formula.
In dimension (threefolds): .
In dimension (fourfolds): ... more precisely, .
The surface case is distinguished by its simplicity: only two Chern numbers ( and ) constrained by one relation (), leaving a one-dimensional "geography."
Summary
Noether's formula is the single most important numerical identity in surface theory. It:
- Constrains invariants: any two of determine the third.
- Underlies Riemann--Roch: the Riemann--Roch theorem for surfaces uses as an input, and Noether's formula expresses it in terms of Chern numbers.
- Bounds geography: combined with the Noether inequality () and BMY inequality (), it delineates the region of possible invariants.
- Classifies surfaces: the four classes of surfaces are distinguished by their Chern numbers.
- Connects topology and algebra: it relates the holomorphic invariant to the topological invariants and .