04.05.03 · algebraic-geometry / divisors

Line bundle on a scheme

shipped3 tiersLean: partial

Anchor (Master): Hartshorne §II; Vakil; Cartan-Serre FAC 1955; Lazarsfeld *Positivity*

Intuition [Beginner]

A line bundle on a scheme $X$ is a rank-1 vector bundle: a sheaf that locally looks like the structure sheaf $O_{X}$ , but with twisted gluing on overlaps. Equivalently, it is a sheaf that becomes isomorphic to $O_{X}$ on a fine enough open cover, with the gluing data encoded by transition functions valued in $O_{X}^{\times}$ .

Line bundles are the algebraic-geometric counterpart of complex line bundles on a topological space, except that local freeness is in the Zariski topology and the transition functions are algebraic (regular invertible functions). They are everywhere in algebraic geometry: every divisor gives a line bundle (and conversely, on locally factorial schemes), every embedding into projective space comes from a line bundle (the very ample one), every cohomological computation involves line bundle twists.

The set of isomorphism classes of line bundles on $X$ forms an abelian group under tensor product — the Picard group $Pic (X)$ . This group is the algebraic-geometric measure of "how non-free topologically" $X$ is.

Visual [Beginner]

A scheme covered by trivialising open subsets, with rank-1 transitions encoded by invertible regular functions.

Worked example [Beginner]

The structure-sheaf line bundle $O_{X}$ on any scheme $X$ : the structure sheaf itself, with constant transition function 1.

The twisting sheaf $O (1)$ on $P^{1} = Proj (k [x, y])$ : free on each affine $D (x)$ and $D (y)$ , with transition function $y / x$ on the overlap. As an abstract line bundle, $O (1)$ is the dual of the tautological line bundle on $P^{1}$ — its global sections are the linear forms $a x + b y$ , dimension 2.

More generally on $P^{n}$ , $O (d)$ is a line bundle with transition functions $(x_{i} / x_{j})^{d}$ on $D (x_{i}) \cap D (x_{j})$ . Its global sections are homogeneous polynomials of degree $d$ , dimension $(n n + d)$ .

For an elliptic curve $E$ with origin $O$ : every degree-0 line bundle is $O ([P] - [O])$ for a unique $P \in E$ , and these form an abelian group under tensor product isomorphic to $E$ itself. The degree- $d$ line bundles are translates of these by $O (d \cdot O)$ , giving $Pic (E) ≅ E \times Z$ .

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $X$ be a scheme. A line bundle (synonymously: invertible sheaf) on $X$ is a coherent $O_{X}$ -module $L$ that is locally free of rank 1: every point $x \in X$ has an open neighbourhood $U$ such that $L ∣_{U} ≅ O_{U}$ .

Equivalent characterisations.

(L1) $L$ is a coherent $O_{X}$ -module, and there exists an open cover ${U_{i}}$ of $X$ with $L ∣_{U_{i}} ≅ O_{U_{i}}$ for each $i$ .

(L2) $L$ is locally generated by one element, and the local relations are satisfied by 0 alone (i.e., locally a free rank-1 module).

(L3) The dual $L^{- 1} := H o m_{O_{X}} (L, O_{X})$ satisfies $L \otimes L^{- 1} ≅ O_{X}$ — i.e., $L$ is invertible under tensor product.

Transition functions. Given a trivialising cover ${U_{i}}$ with isomorphisms $φ_{i} : L ∣_{U_{i}} \sim O_{U_{i}}$ , the transition functions are

g_{ij} := φ_{i} \circ φ_{j}^{- 1} : O_{U_{i} \cap U_{j}} \sim O_{U_{i} \cap U_{j}},

multiplicatively given by an element $g_{ij} \in O^{\times} (U_{i} \cap U_{j})$ . The cocycle condition

g_{ij} g_{j k} = g_{ik} on U_{i} \cap U_{j} \cap U_{k}

ensures consistency. Two cocycles ${g_{ij}}, {g_{ij}^{'}}$ define isomorphic line bundles iff they differ by a coboundary: $g_{ij}^{'} = g_{ij} \cdot (s_{j} / s_{i})$ for some $s_{i} \in O^{\times} (U_{i})$ . So line bundles up to isomorphism correspond to elements of first Čech cohomology

Pic (X) = \overset{ˇ}{H}^{1} (X; O_{X}^{\times}) .

Operations.

Tensor product: $L \otimes_{O} M$ is again a line bundle, with transition functions ${g_{ij}^{L} \cdot g_{ij}^{M}}$ .
Dual: $L^{- 1} = H o m (L, O)$ has transition functions ${g_{ij}^{- 1}}$ .
Pullback: for $f : X \to Y$ and $L$ on $Y$ , $f^{*} L$ is a line bundle on $X$ with transition functions ${g_{ij} \circ f}$ .
Tensor power: $L^{\otimes d}$ for $d \in Z$ — positive, negative, or zero.

The Picard group. $Pic (X) := {line bundles on X} / ≅$ , with $\otimes$ as the group operation, $O_{X}$ as identity, $L \mapsto L^{- 1}$ as inverse. As noted above, $Pic (X) ≅ H^{1} (X; O_{X}^{\times})$ .

Connection to Weil divisors. For a locally factorial Noetherian integral scheme $X$ (every local ring is a UFD), there is a natural isomorphism

Pic (X) ≅ Cl (X)

between line bundles up to isomorphism and Weil divisor classes. The map sends a Cartier divisor $D$ to $O (D)$ — the line bundle whose sections are rational functions $f$ with $div (f) + D \geq 0$ .

Examples.

Structure-sheaf line bundle $O_{X}$ : identity in $Pic (X)$ .
Twisting sheaves $O (d)$ on $P^{n}$ : $Pic (P^{n}) = Z$ , generated by $O (1)$ .
Line bundles on an elliptic curve $E$ : $Pic (E) ≅ E \times Z$ .
Line bundles on $Spec (O_{K})$ for number ring $K$ : ideal class group $Cl (O_{K})$ .
Line bundles on smooth projective curve $C$ of genus $g$ : $Pic (C)$ has degree map onto $Z$ with kernel $Jac (C)$ ( $g$ -dim abelian variety).

Key theorem with proof [Intermediate+]

Theorem ( $Pic (X) ≅ H^{1} (X; O_{X}^{\times})$ via Čech cohomology). For any scheme $X$ , the set of isomorphism classes of line bundles forms an abelian group under tensor product, and is naturally isomorphic to $\overset{ˇ}{H}^{1} (X; O_{X}^{\times})$ .

Proof sketch.

From line bundle to cohomology class. Choose a trivialising cover $U = {U_{i}}$ for $L$ , with isomorphisms $φ_{i} : L ∣_{U_{i}} \sim O_{U_{i}}$ . The transition functions $g_{ij} = φ_{i} φ_{j}^{- 1} \in O^{\times} (U_{i} \cap U_{j})$ form a Čech 1-cocycle: $g_{ij} g_{j k} = g_{ik}$ on triple intersections.

A different choice of trivialisation $φ_{i}^{'} = s_{i} φ_{i}$ gives transition functions $g_{ij}^{'} = (s_{j} / s_{i}) g_{ij}$ — differing by a coboundary. So line bundle isomorphism classes inject into the first Čech cohomology of $O_{X}^{\times}$ over the cover.

From cohomology class to line bundle. Given a Čech 1-cocycle ${g_{ij}}$ , define $L$ on $U$ by $L ∣_{U_{i}} = O_{U_{i}}$ and gluing $L ∣_{U_{i} \cap U_{j}}$ to $L ∣_{U_{j} \cap U_{i}}$ via multiplication by $g_{ij}$ . Cocycle condition makes the gluing consistent on triple intersections.

Tensor product = sum in cohomology. The tensor product $L \otimes M$ has transition functions $g_{ij}^{L} \cdot g_{ij}^{M}$ , which is the sum in the multiplicatively-written abelian group $O_{X}^{\times}$ (so addition in the cohomology).

Refining covers. Taking the colimit over refinements gives Čech cohomology $\overset{ˇ}{H}^{1} (X; O_{X}^{\times})$ . For "reasonable" schemes (e.g., separated quasi-compact), $\overset{ˇ}{H}^{1} = H^{1}$ (sheaf cohomology). $□$

This isomorphism is the foundational identification: line bundles are exactly $H^{1}$ of the multiplicative structure sheaf. It links the geometric notion (line bundle) to the cohomological invariant (Picard group as $H^{1}$ ).

Bridge. The construction here builds toward 04.05.02 (picard group), where the same data is upgraded, and the symmetry side is taken up in 04.05.05 (ample and very ample line bundle). The defining pattern appears again in those units in a sharpened form, where the local data is glued or quotiented. Putting these together, the foundational insight is that the data of this unit gives the structural signature that the rest of the strand reads off.

Exercises [Intermediate+]

Exercise 5 (medium, conceptual).

Why is $Pic (X) = H^{1} (X; O_{X}^{\times})$ as a sheaf-cohomological identity?

Hint

Line bundles up to isomorphism correspond to Čech 1-cocycles up to coboundary.

Answer

A line bundle is determined up to isomorphism by its system of transition functions ${g_{ij}}$ — Čech 1-cocycles in the multiplicative structure sheaf $O_{X}^{\times}$ . Two cocycles giving isomorphic line bundles differ by a coboundary ${s_{i}}$ , $g_{ij}^{'} = (s_{j} / s_{i}) g_{ij}$ .

So ${line bundles} / ≅$ corresponds to ${\overset{ˇ}{C} ech 1-cocycles} / {coboundaries} = \overset{ˇ}{H}^{1} (X; O_{X}^{\times})$ .

For "reasonable" schemes — separated, quasi-compact — Čech and sheaf cohomology agree, so $Pic (X) ≅ H^{1} (X; O_{X}^{\times})$ as a sheaf-cohomological identity.

This is foundational: it means line bundles are classified by an abelian-group-valued cohomology theory, just as topological line bundles on a topological space $Y$ are classified by $H^{1} (Y; Z /2)$ (over $R$ ) or $H^{2} (Y; Z)$ (Chern class, over $C$ ). The algebraic-geometric and topological pictures align cleanly.

Exercise 6 (hard, proof).

Show the exponential sequence for a complex projective variety $X$ :

0 \to Z \to O_{X} e x p O_{X}^{\times} \to 0,

and use it to compute $Pic (X)$ via long exact sequence.

Hint

Apply $exp (2 π i \cdot \cdot)$ to a holomorphic function.

Answer

For $X$ a smooth complex projective variety, the holomorphic exponential gives a short exact sequence of analytic sheaves

0 \to \underline{Z} \to O_{X} e x p (2 π i \cdot) O_{X}^{\times} \to 0.

The map $exp$ has kernel exactly the constant sheaf $\underline{Z}$ (since $exp$ is locally surjective and has $Z$ -valued kernel).

Long exact sequence in cohomology:

\dots \to H^{1} (X; Z) \to H^{1} (X; O_{X}) \to H^{1} (X; O_{X}^{\times}) \to H^{2} (X; Z) \to H^{2} (X; O_{X}) \to \dots

Identifying $H^{1} (X; O_{X}^{\times}) = Pic (X)$ :

Pic (X) \to H^{2} (X; Z) c_{1} H^{2} (X; O_{X})

The first map sends a line bundle to its first Chern class $c_{1} (L) \in H^{2} (X; Z)$ . The image is exactly the kernel of the next map — the Néron-Severi group $NS (X) \subseteq H^{2} (X; Z)$ of $(1, 1)$ -classes (Lefschetz $(1, 1)$ -theorem).

The kernel of $Pic (X) \to NS (X)$ is the Picard variety $Pic^{0} (X)$ , an abelian variety.

So $Pic (X)$ fits into the exact sequence

0 \to Pic^{0} (X) \to Pic (X) \to NS (X) \to 0,

with $Pic^{0} (X) = H^{1} (X; O) / H^{1} (X; Z)$ a complex torus (the Albanese variety dual). This decomposition is one of the deepest structural facts about complex algebraic varieties.

Exercise 7 (hard, conceptual).

State and explain the first Chern class of a line bundle.

Hint

The first Chern class is the cohomological invariant of a line bundle in $H^{2}$ .

Answer

For a complex line bundle $L$ on a topological space (or smooth manifold) $X$ , the first Chern class is

c_{1} (L) \in H^{2} (X; Z) .

In algebraic-geometric terms, it is the connecting homomorphism in the exponential sequence (Exercise 6).

Properties:

Naturality: $c_{1} (f^{*} L) = f^{*} c_{1} (L)$ .
Tensor product: $c_{1} (L \otimes M) = c_{1} (L) + c_{1} (M)$ — so $c_{1}$ is a homomorphism $Pic (X) \to H^{2} (X; Z)$ .
Integrality: $c_{1} (L) \in H^{2} (X; Z)$ , an integral class.
Geometric content: $c_{1} (O (D)) = [D] \in H^{2}$ is the Poincaré dual of the divisor $D$ — the cycle class of $D$ .

For complex algebraic varieties, the Lefschetz $(1, 1)$ -theorem says $c_{1}$ identifies $Pic (X)$ with the integral $(1, 1)$ -classes in $H^{2} (X; Z)$ — the Néron-Severi group $NS (X)$ .

The first Chern class is the simplest characteristic class and the prototype for higher Chern classes $c_{i} (E) \in H^{2 i} (X; Z)$ for vector bundles $E$ of higher rank. Chern classes are the foundational topological invariants of complex vector bundles, with deep connections to:

Riemann-Roch: Hirzebruch-Riemann-Roch expresses $χ (E)$ in terms of Chern classes.
K-theory: Chern classes give the map $K^{0} (X) \to H^{*} (X; Q)$ .
Index theory: the Atiyah-Singer index theorem expresses analytical indices in terms of Chern classes.
Mirror symmetry: Chern classes appear on both sides of the mirror correspondence.

Lean formalization [Intermediate+]

lean_status: partial — Mathlib has AlgebraicGeometry.LineBundle (in development), invertible-sheaf API, and the corresponding tensor-product structure.

[object Promise]

Advanced results [Master]

Picard scheme. The functor "line bundles relative to a base $S$ " is representable by a scheme $Pic_{X / S}$ , the Picard scheme. Its identity component $Pic_{X / S}^{0}$ is the Picard variety — for smooth projective curves, the Jacobian. Picard schemes are foundational in moduli theory and arithmetic geometry.

Riemann-Roch and Hirzebruch-Riemann-Roch. Riemann-Roch on a smooth projective curve gives $χ (L) = de g L + 1 - g$ for a line bundle $L$ . HRR generalises to higher dimensions: $χ (L) = \int_{X} ch (L) td (X)$ .

Ample, big, nef. Positivity properties of line bundles: ample (sufficiently positive twist embeds the variety), nef (numerically eventually free), big (asymptotic global sections large). These positivity classes drive the minimal model program (Mori, Reid, Birkar) for higher-dimensional varieties.

Hodge bundle. On the moduli space $M_{g}$ of genus- $g$ curves, the Hodge bundle $Λ$ is the rank- $g$ vector bundle with fibre $H^{0} (C; ω_{C})$ at $[C]$ . Its determinant gives a fundamental line bundle on $M_{g}$ , with Mumford's relations linking its powers to other tautological line bundles.

Tate twist and arithmetic. Over a number field, the Tate twist $Q_{ℓ} (n)$ is the pullback of the cyclotomic character. Étale cohomology with Tate twists (e.g., $H_{\overset{e}{ˊ} t}^{i} (X_{\overset{ˉ}{K}}; Q_{ℓ} (n))$ ) carries the Galois representation that controls $L$ -functions. Line bundles on arithmetic schemes feed into the Birch and Swinnerton-Dyer conjecture and modern arithmetic geometry.

Generalised line bundles. Beyond ordinary line bundles: gerbes (a categorified notion of line bundles), 2-gerbes, Brauer-Severi varieties (twisted line-bundle structures), and the higher Brauer group $Br (X) = H^{2} (X; O_{X}^{\times})$ .

Geometric Langlands and line bundles on $Bun_{G}$ . $Bun_{G}$ — the moduli of $G$ -bundles on a curve — has line bundles corresponding to level structures and theta functions, with deep connections to integrable systems and conformal field theory.

Synthesis. This construction generalises the pattern fixed in 04.06.02 (coherent sheaf), with the symmetric data replaced by its skew or twisted analogue. Read in the opposite direction, the construction is dual to the metric story: complements and orthogonality are taken with respect to the bilinear datum of this unit, not a metric, and the resulting category of subobjects is the one the rest of the strand classifies. The central insight is that this datum identifies algebra with geometry: functions become vector fields, subspaces become quotients, and invariants become cohomology classes — and that identification is the engine driving every theorem downstream.

Full proof set [Master]

Detailed proofs of: $Pic (X) = H^{1} (X; O_{X}^{\times})$ at the sheaf-cohomology level, the equivalence of Cartier divisors and line bundles, Lefschetz $(1, 1)$ -theorem, the Picard-scheme representability — these are deferred to companion units. The core formal content is in the formal-definition section.

Connections [Master]

Coherent sheaf 04.06.02 — line bundles are coherent sheaves with locally-free-rank-1 structure.
Quasi-coherent sheaf 04.06.01 — line bundles are coherent (hence quasi-coherent).
Weil divisor 04.05.01 — Weil divisors and line bundles agree via $Pic = Cl$ on locally factorial schemes.
Picard group 04.05.02 — line bundles up to isomorphism form the Picard group.
Cartier divisor 04.05.04 — Cartier divisors and line bundles correspond bijectively.
Ample line bundle 04.05.05 — ampleness is the key positivity notion for line bundles.
Riemann-Roch theorem for curves 04.04.01 — Riemann-Roch is a calculation involving line bundles.
Sheaf cohomology 04.03.01 — cohomology of line bundles drives the geometry.
Projective space 04.07.01 — the universal target for projective embeddings is determined by an ample line bundle.

Historical & philosophical context [Master]

The notion of line bundle in algebraic geometry has a rich history. The classical theory of abelian integrals and theta functions on Riemann surfaces (Riemann 1857, Weierstrass) implicitly used line bundles before the language existed. André Weil's Foundations of Algebraic Geometry (1946) treated divisor classes systematically.

The modern sheaf-theoretic formulation came with Cartan-Serre's Faisceaux Algébriques Cohérents (1955). FAC introduced invertible sheaves on algebraic varieties, with the tensor-product group structure and the cohomological identification $Pic = H^{1} (O^{\times})$ . The twisting sheaves $O (d)$ on projective space were systematically developed.

Grothendieck's EGA (1960–67) extended line bundle theory to schemes:

Functorial line bundle theory over arbitrary base schemes.
Picard scheme (Grothendieck, Bourbaki seminar 1962): the moduli space of line bundles, representable by an algebraic scheme under appropriate hypotheses.
Cartier divisors and the comparison with Weil divisors.
Ample, nef, big line bundles in modern formulation, building on Kodaira and Serre's work.

The 20th century saw line bundles become central:

Riemann-Roch and Hirzebruch-Riemann-Roch (Hirzebruch 1954) compute Euler characteristics in terms of Chern classes of line bundles.
Atiyah-Singer index theorem (1963) generalises to elliptic operators on bundles.
Birkenhake-Lange Complex Abelian Varieties (1992) systematises the Picard variety theory.
Minimal model program (Mori, Kawamata, Reid, Shokurov, Birkar) classifies higher-dimensional varieties via positivity of canonical line bundles.

Today line bundles are ubiquitous: in classification of varieties (canonical line bundles), in moduli theory (Hodge bundles), in arithmetic geometry (Tate modules and $L$ -functions), in mathematical physics (anomalies, gerbes, geometric quantisation), and in the geometric Langlands programme. The simple notion of "rank-1 locally free sheaf" generates an essentially endless mathematical landscape.

Bibliography [Master]

Hartshorne, Algebraic Geometry — §II.6 is the standard introduction.
Vakil, The Rising Sea: Foundations of Algebraic Geometry — §14, modern treatment.
Lazarsfeld, Positivity in Algebraic Geometry I and II — comprehensive treatment of line bundles, ampleness, and positivity.
Eisenbud-Harris, The Geometry of Schemes — Ch. III, geometric perspective.
Mumford, Abelian Varieties — Ch. III, line bundles on abelian varieties.
Birkenhake-Lange, Complex Abelian Varieties — comprehensive line-bundle theory on complex tori.
Cartan-Serre, Faisceaux Algébriques Cohérents (FAC, 1955) — the foundational invertible-sheaf treatment.
Grothendieck-Dieudonné, EGA II, IV — scheme-theoretic foundations.