Castelnuovo's Rationality Criterion
Castelnuovo's rationality criterion is one of the crown jewels of the Italian school of algebraic geometry. It provides a complete numerical characterization of rational surfaces: a surface is rational if and only if its irregularity and second plurigenus both vanish. This result is far deeper than its one-dimensional analogue (Luroth's theorem) and reveals that the birational geometry of surfaces is governed by subtle cohomological invariants.
Background: birational invariants of surfaces
Let be a smooth projective surface over an algebraically closed field . The following are birational invariants (unchanged under birational equivalence):
- The irregularity: , the dimension of the Albanese variety.
- The geometric genus: , the number of holomorphic -forms.
- The -th plurigenus: for . Note .
- The Kodaira dimension: the growth rate of as (values ).
For a rational surface ( birational to ), all these invariants vanish: , for all , and .
A smooth projective surface over is:
- Rational if is birational to , i.e., is a purely transcendental extension.
- Unirational if there exists a dominant rational map , i.e., as a subfield.
Every rational surface is unirational. The converse is the Luroth problem for surfaces.
The main theorem
Let be a smooth projective surface over an algebraically closed field of characteristic . Then is rational if and only if:
Equivalently, is rational if and only if and .
The condition ensures the Albanese variety is trivial (), so has no "abelian variety part." The condition kills the possibility of nontrivial pluricanonical forms. Together they force to be rational.
The condition is strictly stronger than for the purpose of detecting rationality. The second plurigenus is the critical invariant: it distinguishes Enriques surfaces (which have but ) from rational surfaces.
Comparison with Luroth's theorem
Let be a smooth projective curve over an algebraically closed field (any characteristic). If is unirational (i.e., there exists a dominant rational map ), then is rational (i.e., ).
Equivalently: every subfield of containing and of transcendence degree over is itself purely transcendental.
Luroth's theorem says that for curves, unirational = rational. The natural question is whether this extends to higher dimensions:
- Dimension 1: Unirational rational (Luroth, 1876). True in all characteristics.
- Dimension 2, char : Unirational rational (Castelnuovo, 1894). This is a consequence of the rationality criterion.
- Dimension 2, char : Unirational rational in general (Zariski, Shioda). See below.
- Dimension : Unirational rational (Clemens--Griffiths, Iskovskikh--Manin, Artin--Mumford, 1972). See below.
Castelnuovo's theorem is the last dimension where the Luroth-type statement holds in characteristic .
If is unirational, there is a dominant rational map . Then:
- : A dominant map shows that (using the Hodge symmetry in char ).
- : Similarly shows .
So , and Castelnuovo's criterion gives that is rational.
Proof outline
Necessity (): If is rational, then is birational to . Since and are birational invariants, and , for all , we get and .
Sufficiency (): This is the deep direction. Assume and . We must show is rational. The argument proceeds in several steps.
Step 1. Show . Since (because via multiplication by any nonzero section, but actually we just note and a nonzero section of gives a nonzero section of by squaring; but if , we can square a nonzero -form to get a nonzero section of ). More precisely: if is nonzero, then is nonzero, so , contradiction. Hence .
Step 2. Compute . With and : .
Step 3. Find a pencil of rational curves. By Riemann--Roch for surfaces, for any divisor : . Using and the structure theory of surfaces (specifically the Enriques--Kodaira classification), one shows that is nonempty or admits a ruling.
The key argument: since , the Kodaira dimension satisfies . By the classification of surfaces with , is either rational or ruled over a curve of genus . Since , if is ruled, it is ruled over .
Step 4. A ruled surface over is rational. A surface ruled over is birational to , which is rational (birational to ).
Hence is rational.
The proof of sufficiency relies heavily on the Enriques--Kodaira classification of surfaces. The key input is that implies ruled or rational. Without classification theory, the bare conditions , seem insufficient to construct a birational map to .
Historically, Castelnuovo's original proof (1894) predated the full classification and used a more direct approach involving linear systems and the construction of a pencil of rational curves on . The modern proof using is cleaner but conceptually depends on deeper structure theory.
Why alone is not enough
An Enriques surface (in characteristic ) satisfies:
- (no irregular part),
- (no holomorphic -forms),
- (there exists a nonzero section of ),
- but (the canonical bundle is -torsion),
- (Kodaira dimension zero).
Since , Castelnuovo's criterion correctly identifies as not rational. The Enriques surface is the classical counterexample showing that is not sufficient for rationality. One genuinely needs .
Construction: An Enriques surface is the quotient of a K3 surface by a fixed-point-free involution . Since , the involution acts as on , so (i.e., ), but .
A Godeaux surface is a surface of general type with , , and . By the plurigenus formula for surfaces of general type:
So , and Castelnuovo's criterion correctly says the Godeaux surface is not rational. Despite , it has Kodaira dimension .
Construction: The classical Godeaux surface is the quotient , where is the Fermat quintic and acts by with .
Why alone is not enough
Let be a smooth curve of genus , and let . Then:
- (since , by the Kunneth formula ),
- (since ).
When , we have and (since for ruled surfaces), but . The surface is not rational (it is ruled over , and the base is a birational invariant). Castelnuovo's criterion detects this via .
An abelian surface has , , . It is very far from rational. The large irregularity reflects the rich structure of the Albanese variety itself.
Classical examples: rational surfaces
is the prototypical rational surface:
- , , for all (since has no sections, nor does any positive tensor power ).
- .
- .
Castelnuovo's criterion: , rational. β
The Hirzebruch surface is a ruled surface over :
- (base is ), , for all .
- .
- (independent of , since all are birational to ).
All are rational by Castelnuovo's criterion. Concretely: is rational via projection from a point; is rational by construction; for is rational since one can blow up and blow down to reduce to or .
A smooth cubic surface is isomorphic to the blowup of at points in general position, so it is rational. Verifying via Castelnuovo:
- . Since has no global sections, .
- , which also has no global sections. So .
- By the Lefschetz hyperplane theorem, (or directly: is torsion-free and ).
So , rational by Castelnuovo. β This also confirms the classical result that every smooth cubic surface is rational.
A Del Pezzo surface is a smooth projective surface with ample. These are classified: is either or (blowup of at points in general position), with .
Every Del Pezzo surface satisfies:
- : since is ample, is anti-ample, so .
- : since is ample, .
- : from the construction as blowups of , or from .
Castelnuovo's criterion gives: all Del Pezzo surfaces are rational. β
The smooth quadric is a Del Pezzo surface with :
- (by adjunction: , which as a bidegree is under the Segre embedding).
- , .
- .
Rational by Castelnuovo. β The explicit birational map is given by projection from any point on .
Complete intersections
Let be a smooth complete intersection of two quadrics. Then is a Del Pezzo surface of degree with :
- .
- , .
- (Lefschetz).
Rational by Castelnuovo. β This surface is the blowup of at points in general position.
Not all smooth complete intersections are rational. Consider:
- in : . So . This is a K3 surface, not rational.
- in : . Again a K3 surface, , not rational.
- in (quartic surface): . K3 surface, , not rational.
- in (quintic surface): . Surface of general type with , very far from rational.
The boundary for rationality of hypersurfaces in is: degree gives rational, degree gives K3 (not rational), degree gives general type (not rational).
The Enriques surface: the critical counterexample
The Enriques surface is the key example that justifies the need for rather than just :
- , , so .
- (the canonical class is nonzero in ).
- (the canonical class is -torsion in ).
- , but .
- The plurigenera cycle: for odd, for even.
- (Kodaira dimension zero).
- .
- , (the Picard number equals the second Betti number).
The Enriques surface is not rational, not ruled, and not unirational. It has Kodaira dimension , sitting in the classification alongside K3 surfaces, abelian surfaces, and hyperelliptic (bielliptic) surfaces.
Why blocks rationality: If were rational, then . Since , cannot be birational to .
The Barlow surface: an exotic rational-like surface
The Barlow surface (R. Barlow, 1985) is a smooth minimal surface of general type with:
- , , so .
- (same as a Godeaux surface).
- (by the plurigenus formula for general type).
- (simply connected, unlike the classical Godeaux surface which has ).
Despite being simply connected with (sharing these properties with ), the Barlow surface is not rational because . Castelnuovo's criterion detects this.
The Barlow surface was the first known simply connected surface of general type with , answering a long-standing question.
Failure of Luroth in dimension
There exist smooth projective threefolds that are unirational but not rational. Three fundamentally different constructions were found in 1972:
- Clemens--Griffiths: A smooth cubic threefold is unirational but not rational. The obstruction is that the intermediate Jacobian is not a product of Jacobians of curves (it is a principally polarized abelian variety of dimension that is not a Jacobian).
- Iskovskikh--Manin: A smooth quartic threefold is not rational (and conjecturally not even unirational). The proof uses birational rigidity: .
- Artin--Mumford: Certain conic bundles over are unirational but not rational. The obstruction is the Brauer group: while . In fact, .
Let be a smooth cubic hypersurface. Then:
- is unirational: projecting from a line gives a conic bundle over , which is dominated by (after a rational section). Alternatively, the map sending a line through a point on to the residual intersection point gives a dominant map .
- is not rational: the intermediate Jacobian is a principally polarized abelian variety of dimension . If were rational, would be a product of Jacobians of curves. But for a general cubic threefold, is an irreducible ppav (not isomorphic to any Jacobian), contradiction.
This shows that no analogue of Castelnuovo's criterion can hold in dimension using only numerical invariants like plurigenera. Subtler invariants (intermediate Jacobian, Brauer group, etc.) are needed.
Failure in positive characteristic
Over algebraically closed fields of characteristic , there exist unirational surfaces that are not rational. The condition is necessary but not sufficient for rationality in positive characteristic.
Zariski (1958) first observed this phenomenon for certain unirational surfaces in characteristic , and Shioda (1974) gave explicit examples:
- Shioda's examples: In characteristic , the Fermat surface in can be unirational when (or more precisely when certain divisibility conditions hold), even though it is a K3 surface (degree ) or a surface of general type (degree ). For instance, the Fermat quartic is unirational over but is a K3 surface, not rational.
The issue is that inseparable maps (which exist only in char ) can dominate surfaces with nonzero birational invariants, because the pullback on differential forms is zero for purely inseparable morphisms.
The Fermat quartic over :
- is a K3 surface: , , .
- is unirational over : there exists a dominant inseparable rational map .
- is not rational: since , cannot be rational.
This does not contradict Castelnuovo's criterion (which gives , so does not predict rationality). Rather, it shows that the implication "unirational " fails in char due to inseparable maps, and hence the Castelnuovo-style proof that "unirational rational" breaks down.
Surfaces with
Let be a smooth projective surface over an algebraically closed field of characteristic . If (equivalently, for all ), then is ruled: there exists a smooth curve and a birational equivalence .
In particular, . So:
- and is ruled over is rational.
- and is ruled over a curve of genus , not rational.
This theorem shows the logical structure: implies (since for in the Enriques classification). Then implies ruled. Finally, implies the base is , so is rational.
The condition is used rather than for all because already forces : for surfaces, implies (a result from the Enriques classification), and in fact whenever except possibly for some cases, but the only surface with would need with , which does not occur in the classification (K3 has , abelian has , Enriques has , bielliptic has ).
Application: verifying rationality of specific surfaces
Let be the blowup of at points. Since blowup does not change birational class, is automatically birational to , hence rational. Castelnuovo verifies:
- (blowup does not change ).
- (blowup does not change plurigenera).
So , , rational. β
The invariants: , , with intersection form of signature .
A conic bundle is a surface with a morphism to whose generic fiber is a smooth conic. If the generic fiber has a rational point (which is automatic over algebraically closed fields), then is birational to , hence rational.
From Castelnuovo: a conic bundle over is ruled over , so and , confirming rationality. β
If the base is a curve of genus , then , and the surface is ruled but not rational.
The Noether--Enriques criterion and Castelnuovo's lemma
A smooth rational curve on a smooth surface with can be contracted: there exists a morphism to a smooth surface such that and is an isomorphism on .
Equivalently, is a -curve if and only if and .
To apply Castelnuovo's criterion effectively, one often first passes to the minimal model of : successively contract all -curves. Since and are birational invariants, they are unchanged.
For a rational surface, the minimal model is either or (). So verifying rationality via Castelnuovo reduces to checking , then the classification tells us the minimal model is one of these standard rational surfaces.
The Enriques--Kodaira classification and rationality
The Enriques--Kodaira classification organizes surfaces by Kodaira dimension :
(ruled surfaces): for all . Rational iff .
- Rational surfaces (): , , blowups.
- Irrational ruled (): ruled over curves of genus .
: is bounded, . Never rational. Includes:
- K3 surfaces: , , .
- Enriques surfaces: , , .
- Abelian surfaces: , , .
- Bielliptic (hyperelliptic) surfaces: , , .
(properly elliptic surfaces): grows linearly. Never rational.
(surfaces of general type): grows quadratically. Never rational.
Castelnuovo's criterion captures exactly the , cell in this table.
Numerical invariants of classical surfaces
A summary of the invariants that Castelnuovo's criterion uses:
Rational surfaces (, , rational β):
- : , , , .
- : , , , .
- Cubic surface in : , , , .
- Del Pezzo degree : , , , .
Non-rational with ():
- K3 surface: , , , .
- Enriques surface: , , , .
- Godeaux surface: , , , .
- Barlow surface: , , , .
Non-rational with :
- Ruled over genus- curve: , , , .
- Abelian surface: , , , .
The role of the Albanese variety
For a smooth projective variety over , the Albanese variety is , an abelian variety of dimension .
The Albanese map is universal among maps from to abelian varieties.
For surfaces:
- : , the Albanese map is constant.
- : is an elliptic curve, and is a fibration (for minimal surfaces of ).
- : is an abelian surface, and can be birational (for abelian surfaces), or a fibration.
For a rational surface :
since .
The condition in Castelnuovo's criterion says precisely that : the surface has no "abelian variety component." Any dominant map from to an abelian variety is constant.
For a ruled surface with , we have , detecting that is not rational even though all plurigenera vanish.
Advanced example: rationality of specific surfaces
The del Pezzo surface of degree is with four points in general position. It embeds as a surface of degree in via the anticanonical linear system .
- , , : rational by Castelnuovo. β
- , .
- (the symmetric group on letters).
- Contains exactly lines (the -curves).
This surface has the interesting property that it is the unique del Pezzo surface of degree up to isomorphism (unlike degrees where moduli exist).
The del Pezzo surface of degree is with three non-collinear points:
- , , : rational by Castelnuovo. β
- .
- Contains exactly lines (the three exceptional divisors and the three strict transforms of lines ).
- is also the blowup of at two points.
- is a toric variety, corresponding to the hexagonal fan.
Historical context and significance
The story of Castelnuovo's rationality criterion spans over a century:
1876: Luroth proves that unirational curves are rational.
1894: Castelnuovo proves that for surfaces over , implies rationality. As a corollary, unirational surfaces are rational in characteristic .
1914--1949: The Italian school (Enriques, Castelnuovo, Severi) develops the classification of surfaces. Enriques surfaces are recognized as the key examples with but .
1958: Zariski discovers that Castelnuovo's theorem can fail in positive characteristic, finding unirational surfaces that are not rational.
1971--1972: Clemens--Griffiths, Iskovskikh--Manin, and Artin--Mumford independently show that Luroth's theorem fails in dimension , using three different methods (intermediate Jacobians, birational rigidity, Brauer groups).
1974: Shioda gives explicit examples of unirational K3 surfaces in positive characteristic.
1985: Barlow constructs a simply connected surface of general type with , showing that the fundamental group does not determine rationality.
Summary
Castelnuovo's rationality criterion is remarkable for several reasons:
-
Numerical characterization: Rationality, a priori a birational property requiring the construction of an explicit map to , is completely detected by two numerical invariants and .
-
Optimal conditions: The conditions and are both necessary and jointly sufficient. Neither alone suffices: Enriques surfaces have but ; ruled surfaces over curves of positive genus have but .
-
Resolution of Luroth for surfaces: It implies that every unirational surface over is rational, settling the Luroth problem in dimension .
-
Characteristic sensitivity: The theorem holds in characteristic but can fail in characteristic due to inseparable maps, highlighting the role of separability in birational geometry.
-
Dimensional boundary: In dimension , no such purely numerical criterion exists, and detecting rationality requires more subtle invariants (intermediate Jacobians, Brauer groups, derived categories, etc.).
-
Foundation of classification: The criterion is intimately connected to the Enriques--Kodaira classification and illustrates how the interplay of , , and plurigenera organizes the birational geometry of surfaces.