K3 surface

K3 surface

Dans la seconde partie de mon rapport, il s'agit des variétés kählériennes dites K3, ainsi nommées en l'honneur de Kummer, Kähler, Kodaira et de la belle montagne K2 au Cachemire

right, André Weil (1958, p.546), describing the reason for the name "K3 surface"

In mathematics, a K3 surface is a complex or algebraic smooth minimal complete surface that is regular and has trivial canonical bundle.

In the Enriques–Kodaira classification of surfaces they form one of the 4 classes of surfaces of Kodaira dimension 0.

Together with two-dimensional complex tori, they are the Calabi–Yau manifolds of dimension two. Most complex K3 surfaces are not algebraic. This means that they cannot be embedded in any projective space as a surface defined by polynomial equations. André Weil (1958) named them in honor of three algebraic geometers, Kummer, Kähler and Kodaira, and the mountain K2 in Kashmir.


  • Definition 1
  • Calculation of the Betti numbers 2
  • Properties 3
  • The period map 4
  • Projective K3 surfaces 5
  • Relation to string duality 6
  • Examples 7
  • See also 8
  • References 9
  • External links 10


There are many equivalent properties that can be used to characterize a K3 surface. The only complete smooth surfaces with trivial canonical bundle are K3 surfaces and tori (or abelian varieties), so one can add any condition to exclude the latter to define K3 surfaces. Over the complex numbers the condition that the surface is simply connected is sometimes used.

There are a few variations of the definition: some authors restrict to projective surfaces, and some allow surfaces with Du Val singularities.

Calculation of the Betti numbers

Equivalently to the above definition, K3 surface S is defined as the surface that has trivial canonical bundle KS = 0 and irregularity q = 0. Then there is a trivial mapping from S to P1 and q = h^{0,1} = \text{dim} H^1(S,\mathcal{O}_S) =0.

From Serre duality


Combining with this, one obtains the Euler characteristic

\chi(S,\mathcal{O}_S):=\sum_i (-1)^i h^i(S,\mathcal{O}_S)=2.

On the other hand, from the Riemann-Roch Theorem (Noether's formula)


where ci denotes i-th Chern class. KS is trivial and so the first Chern class c1 = 0. The Euler number e(S) is equal to the second Chern class c2(S) and then e(S) = 24. Therefore, b1 = 0, b2 = 22.


1. All complex K3 surfaces are diffeomorphic to one another (proved by Kunihiko Kodaira firstly). Siu (1983) showed that all complex K3 surfaces are Kähler manifolds. As a consequence of this and Yau's solution to the Calabi conjecture, all complex K3 surfaces admit Ricci-flat metrics.

2. The (p,q)-th cohomology group is well known. The Hodge diamond is

0 0
1 20 1
0 0

3. On H^2(S,\mathbb{Z}) this defines a lattice structure, called the K3 lattice, as described in the next section.

Because of the above properties, K3 surfaces have been studied extensively not only in algebraic geometry but also in Kac–Moody algebras, mirror symmetry and string theory. In particular, the lattice structure provides the modularity with the Néron–Severi group on it.

The period map

There is a coarse moduli space for marked complex K3 surfaces, a non-Hausdorff smooth analytic space of complex dimension 20. There is a period mapping and the Torelli theorem holds for complex K3 surfaces.

If M is the set of pairs consisting of a complex K3 surface S and a Kähler class of H1,1(S,R) then M is in a natural way a real analytic manifold of dimension 60. There is a refined period map from M to a space KΩ0 that is an isomorphism. The space of periods can be described explicitly as follows:

  • L is the even unimodular lattice II3,19.
  • Ω is the Hermitian symmetric space consisting of the elements of the complex projective space of LC that are represented by elements ω with (ω,ω)=0, (ω,ω^*)>0.
  • KΩ is the set of pairs (κ, [ω]) in (LR, Ω) with (κ,E(ω))=0, (κ,κ)>0.
  • 0 is the set of elements (κ, [ω]) of KΩ such that (κd) ≠ 0 for every d in L with (d,d)=−2, (ω,d)=0.

Projective K3 surfaces

If L is a line bundle on a K3 surface, then the curves in the linear system have genus g where c12(L) =2g-2. A K3 surface with a line bundle L like this is called a K3 surface of genus g. A K3 surface may have many different line bundles making it into a K3 surface of genus g for many different values of g. The space of sections of the line bundle has dimension g+1, so there is a morphism of the K3 surface to projective space of dimension g. There is a moduli space Fg of K3 surfaces with a primitive ample line bundle L with c12(L) =2g-2, which is nonempty of dimension 19 for g≥ 2. Mukai (2006) showed that this moduli space Fg is unirational if g≤13, and V. A. Gritsenko, Klaus Hulek, and G. K. Sankaran (2007) showed that it is of general type if g≥63. Voisin (2008) gave a survey of this area.

Relation to string duality

K3 surfaces appear almost ubiquitously in string duality and provide an important tool for the understanding of it. String compactifications on these surfaces are not trivial, yet they are simple enough for us to analyze most of their properties in detail. The type IIA string, the type IIB string, the E8×E8 heterotic string, the Spin(32)/Z2 heterotic string, and M-theory are related by compactification on a K3 surface. For example, the Type IIA compactified on a K3 surface is equivalent to the heterotic string compactified on 4-torus Aspinwall (1996).


  • A double cover of the projective plane branched along a non-singular degree 6 curve is a genus 2 K3 surface.
  • A Kummer surface is the quotient of a two-dimensional abelian variety A by the action a → −a. This results in 16 singularities, at the 2-torsion points of A. The minimal resolution of this quotient is a genus 3 K3 surface.
  • A non-singular degree 4 surface in P3 is a genus 3 K3 surface.
  • The intersection of a quadric and a cubic in P4 gives genus 4 K3 surfaces.
  • The intersection of three quadrics in P5 gives genus 5 K3 surfaces.
  • Brown (2007) describes a computer database of K3 surfaces.

See also


  • Barth, Wolf P.; Hulek, Klaus; Peters, Chris A.M.; Van de Ven, Antonius (2004), Compact Complex Surfaces, Berlin: Springer,  
  • Beauville, Arnaud (1983), "Surfaces K3", Bourbaki seminar, Vol. 1982/83 Exp 609, Astérisque 105, Paris:  
  • Beauville, A.; Bourguignon, J.-P.; Demazure, M. (1985), Géométrie des surfaces K3: modules et périodes, Séminaires Palaiseau, Astérisque 126, Paris:  
  • Brown, Gavin (2007), "A database of polarized K3 surfaces", Experimental Mathematics 16 (1): 7–20,  
  • Burns, Dan; Rapoport, Michael (1975), "On the Torelli problem for kählerian K-3 surfaces", Annales Scientifiques de l'École Normale Supérieure. Quatrième Série 8 (2): 235–273,  
  • Gritsenko, V. A.; Hulek, Klaus; Sankaran, G. K. (2007), "The Kodaira dimension of the moduli of K3 surfaces",  
  • Rudakov, A.N. (2001), "K3 surface", in Hazewinkel, Michiel,  
  • Siu, Y. T. (1983), "Every K3 surface is Kähler",  
  • Voisin, Claire (2008), "Géométrie des espaces de modules de courbes et de surfaces K3 (d'après Gritsenko-Hulek-Sankaran, Farkas-Popa, Mukai, Verra, et al.)" (PDF), Astérisque, Séminaire Bourbaki. 2006/2007. Exp 981 (317): 467–490,  
  • Aspinwall, Paul (1996), K3 Surfaces and String Duality,  

External links

  • Graded Ring Database homepage for a catalog of K3 surfaces
  • The Geometry of K3 surfaces, by David Morrison
  • K3 database for the Magma computer algebra system