Geometric Picard Group
- Geometric Picard Group is the group of isomorphism classes of line bundles modulo algebraic equivalence on varieties or stacks, reflecting both algebraic and topological structures.
- It plays a key role in moduli theory, synthetic algebraic geometry, and arithmetic geometry, with computations based on étale cohomology and tensor power techniques.
- Computational methods combine cohomological invariants, gerbe structures, and explicit divisor class generators to derive ranks, torsion phenomena, and motivic equivalences.
A geometric Picard group is the group of isomorphism classes of line bundles—equivalently, Cartier divisors modulo algebraic equivalence—on an algebraic variety or stack after base change to an algebraic closure. This group encodes the algebraic and topological structure of the underlying object and plays a central role in moduli theory, synthetic algebraic geometry, motivic categories, and arithmetic geometry. Its computation, properties, and classification are subject to deep structural theorems connecting cohomological, stack-theoretic, and motivic perspectives.
1. Definitions and Fundamental Structures
Let be a scheme or stack over a field , and a separable closure. The geometric Picard group is defined as
where divisors are taken modulo algebraic equivalence. For K3 surfaces, is a free -module of rank , equipped with a bilinear intersection form of signature (Elsenhans et al., 2010, Elsenhans et al., 2010). More generally, for projective -space, , with Serre's twisting sheaf as a generator (Cherubini et al., 2024).
The geometric Picard group is always functorial in finite étale morphisms and reflects the global structure after extension of scalars.
2. Synthetic and Stack-Theoretic Computation
In synthetic algebraic geometry, within homotopy type theory (HoTT) and univalence, the Picard group is computed via explicit type-theoretic models. For projective -space, one constructs
and defines line bundles as maps , with the delooping of the unit group . Serre's twisting bundles are defined by explicit tensor powers, and one proves
with every line bundle uniquely expressed as a tensor power of (Cherubini et al., 2024).
For moduli stacks such as the stack of -pointed smooth cyclic covers of the projective line, the geometric Picard group is computed via quotient-stack presentations and equivariant Chow groups. The main theorem provides (Landi, 2023): with explicit geometric generators arising from boundary divisor classes and ramification loci.
Moduli stacks of Jacobians, such as the universal compactified Jacobian , have Picard group decompositions controlled by tautological classes and boundary divisor classes, with connecting exact sequences and gerbe structure determining torsion phenomena (Melo et al., 2010).
3. Cohomological and Motivic Interpretations
In cohomological terms, the geometric Picard group often embeds as a direct summand of étale cohomology via the first Chern class: This facilitates the analysis of rank and intersection properties using Frobenius eigenvalues and Galois module structure, especially for K3 surfaces, where one can compute via point counting and discriminant analysis (Elsenhans et al., 2010, Elsenhans et al., 2010).
For the motivic category , the geometric Picard subgroup generated by affine quadrics is described via reduced motives and determinant classes. Vishik's results yield canonical injections
where runs over indecomposable anisotropic summands, and relations correspond precisely to isomorphisms in classical Chow motives (Vishik, 2018). The Grothendieck–Witt ring embeds into the Picard group via classes of quadratic forms.
4. Computational Techniques and Algorithms
For K3 surfaces over , new methods establish Picard rank one using either reduction at a single prime or, refined, using two primes together with Galois module analysis. The main criteria involve verifying:
- The rank bound from Frobenius eigenvalues.
- Absence of liftable "extra" line bundles via obstruction analysis in deformation theory.
An explicit algorithm, relying on point counting and Artin–Tate formula discriminant computations, concludes if admissible submodules intersect trivially in dimension $2$ or have incompatible discriminants (Elsenhans et al., 2010, Elsenhans et al., 2010).
5. Torsion, Exact Sequences, and Gerbe Structures
Gerbe rigidification plays a fundamental role in the geometric Picard group of moduli stacks. For the universal Jacobian stack, the presence of a finite -band gerbe introduces torsion dictated by the order , leading to exact sequences: and corresponding torsion quotients (Melo et al., 2010). The existence, or failure, of generalized Poincaré bundles is characterized precisely by this obstruction.
6. Generators, Relations, and Explicit Presentations
Explicit descriptions of generators and relations are central in geometric Picard group calculations. For moduli stacks, generators correspond to boundary-divisor loci, ramification divisors, and Hodge classes, with relations imposed by geometric collision and ramification constraints (Landi, 2023). In the motivic setting, determinants of summands and quadratic form equivalence generate relations vital for motivic equivalence classification (Vishik, 2018).
For projective spaces, the relation
is realized via Serre's twisting sheaves, with uniqueness up to tensor power and cocycle classification in affine covers (Cherubini et al., 2024).
7. Connections with Classical, Synthetic, and Motivic Theory
The geometric Picard group serves as a unifying invariant across classical algebraic geometry, synthetic approaches, and motivic homotopy theory. Synthetic proofs often replace matrix patching arguments with high-level fixed-point and truncation methods, leveraging the homotopy-theoretic structure and univalence. Classical divisor class groups and Chow groups of coarse moduli schemes correspond directly under pull-back to stack-theoretic Picard groups, with gerbe and boundary phenomena marking the precise distinctions.
In summary, geometric Picard group theory encapsulates the topological and arithmetic structure of algebraic and stack-theoretic objects, enabling a granular understanding of line bundle classification, cohomological invariants, and motivic equivalence, as demonstrated across foundational results in synthetic, cohomological, and computational contexts (Cherubini et al., 2024, Elsenhans et al., 2010, Elsenhans et al., 2010, Melo et al., 2010, Vishik, 2018, Landi, 2023).