The construction of a scheme structure on (the
representable functor version of) the Picard group, the
Picard scheme, is an important step in algebraic geometry, in particular in the
duality theory of abelian varieties. It was constructed by , and also described by and . In the cases of most importance to classical algebraic geometry, for a
non-singular complete variety V over a
field of
characteristic zero, the
connected component of the identity in the Picard scheme is an
abelian variety called the
Picard variety and denoted Pic0(
V). The dual of the Picard variety is the
Albanese variety, and in the particular case where
V is a curve, the Picard variety is
naturally isomorphic to the
Jacobian variety of
V. For fields of positive characteristic however,
Igusa constructed an example of a smooth projective surface
S with Pic0(
S) non-reduced, and hence not an
abelian variety. The quotient Pic(
V)/Pic0(
V) is a
finitely-generated abelian group denoted NS(
V), the
Néron–Severi group of
V. In other words, the Picard group fits into an
exact sequence :1\to \mathrm{Pic}^0(V)\to\mathrm{Pic}(V)\to \mathrm{NS}(V)\to 1.\, The fact that the rank of NS(
V) is finite is
Francesco Severi's
theorem of the base; the rank is the
Picard number of
V, often denoted ρ(
V). Geometrically NS(
V) describes the
algebraic equivalence classes of
divisors on
V; that is, using a stronger, non-linear
equivalence relation in place of
linear equivalence of divisors, the classification becomes amenable to discrete invariants. Algebraic equivalence is closely related to
numerical equivalence, an essentially topological classification by
intersection numbers. == Relative Picard scheme ==