06.10.21 · riemann-surfaces / several-complex-variables

Cartan Theorems A and B in $C^{n}$ (with proof)

shipped3 tiersLean: none

Anchor (Master): Cartan, Séminaire ENS 1951–53 (originator); Gunning *Introduction to Holomorphic Functions of Several Variables, Vol. III: Homological Theory* Ch. on Theorems A and B; Grauert–Remmert *Theory of Stein Spaces* §IV–§V

Intuition Beginner

In one complex variable, you can always patch local data into global data on a nice region. If you know what a function should look like near each point — say, where it has poles, or how it twists — you can usually find a single function on the whole region that matches. The Cousin problems ask whether the same patching works in several variables, where the geometry is far richer and the room to move is much larger.

Cartan's two theorems answer this for a special class of regions called Stein regions, of which a polydisc or any domain of holomorphy is an example. Theorem A says global functions are abundant: every local behaviour you could ask for at a point is already realised by some function defined on the whole region. Theorem B says there are no hidden obstructions: the patching always succeeds, with nothing left over to block it.

The payoff is concrete. On a Stein region you can prescribe poles freely and find one global meromorphic function with exactly those poles, and you can prescribe zeros freely and find one global holomorphic function with exactly those zeros.

Visual Beginner

Picture a polydisc in two complex variables as a fat square region. Over it sit two kinds of data: little patches each carrying a local recipe (a pole here, a zero there), and the global functions that try to honour all the recipes at once. Theorem A draws arrows from the global pool down to every patch, showing each local recipe is hit. Theorem B colours the whole obstruction space empty: nothing blocks the assembly.

Worked example Beginner

Take the bidisc $Ω = {(z, w) : ∣ z ∣ < 1, ∣ w ∣ < 1}$ in two variables, a basic Stein region. Ask for a meromorphic function on $Ω$ whose only poles are simple, located along the line $z = 0$ , with residue behaviour matching $1/ z$ there.

Near the line $z = 0$ the recipe is the local function $1/ z$ . Away from that line, say near a point with $z$ not zero, the recipe is just $0$ — no pole wanted. These two local recipes agree on their overlap up to a holomorphic correction, since their difference $1/ z - 0 = 1/ z$ is holomorphic wherever $z$ is non-zero.

The global answer is the single function $f (z, w) = 1/ z$ . It is meromorphic on the whole bidisc, holomorphic away from $z = 0$ , and has exactly the prescribed simple pole along $z = 0$ .

What this tells us: the local pole data assembled into one global meromorphic function with no leftover obstruction. That is the additive Cousin problem solved, and on a Stein region it always works.

Check your understanding Beginner

Formal definition Intermediate+

Let $Ω \subseteq C^{n}$ be open and let $_{n} O$ denote the sheaf of germs of holomorphic functions on $C^{n}$ , restricted to $Ω$ . A sheaf $F$ of $_{n} O ∣_{Ω}$ -modules is coherent when it is locally finitely generated and the kernel of every local presentation map $_{n} O^{p} \to F$ is itself locally finitely generated. Oka's coherence theorem 06.10.20 supplies the foundational instance: $_{n} O$ is coherent over itself, so the category of coherent analytic sheaves is non-empty and closed under finite presentations, kernels, cokernels, and extensions.

The region $Ω$ is a domain of holomorphy, equivalently holomorphically convex in the sense of 06.10.01, equivalently pseudoconvex by the solution of the Levi problem 06.10.05; for open subsets of $C^{n}$ these three conditions coincide, and such an $Ω$ is called Stein. The cohomology $H^{q} (Ω, F)$ is the $q$ -th sheaf (Čech or derived-functor) cohomology of $Ω$ with coefficients in $F$ ; the two agree because coherent analytic sheaves are acyclic on small polydiscs.

Definition (Theorem A and Theorem B as properties). Let $Ω$ be Stein and $F$ coherent on $Ω$ .

$F$ satisfies Theorem A when, for every $x \in Ω$ , the evaluation map $H^{0} (Ω, F) \to F_{x} / m_{x} F_{x}$ from global sections to the stalk modulo the maximal ideal $m_{x}$ of $_{n} O_{x}$ is surjective; equivalently the images of global sections generate $F_{x}$ as an $_{n} O_{x}$ -module.
$F$ satisfies Theorem B when $H^{q} (Ω, F) = 0$ for all $q \geq 1$ .

The content of Cartan's theorems is that every coherent analytic $F$ on every Stein $Ω$ satisfies both A and B. The structure sheaf $_{n} O$ itself is the base case, and its Theorem B is the cohomological reading of the $\overset{ˉ}{\partial}$ -solvability of 06.10.04.

Counterexamples to common slips

Theorem B fails on compact complex manifolds: on the projective line $P^{1}$ one has $H^{1} (P^{1}, O (- 2)) \neq = 0$ , so Steinness is essential, not a convenience.
Coherence is essential: a non-coherent sheaf of $_{n} O$ -modules can carry higher cohomology even on a polydisc, so the local-finite-presentation hypothesis is load-bearing.
Theorem A is about generating the stalk, not about injectivity: global sections may have relations among them; surjectivity onto each stalk is the claim, and it is strictly weaker than freeness.

Key theorem with proof Intermediate+

Theorem (Cartan A and B for Stein open sets in $C^{n}$ ). Let $Ω \subseteq C^{n}$ be a domain of holomorphy and $F$ a coherent analytic sheaf on $Ω$ . Then $H^{q} (Ω, F) = 0$ for all $q \geq 1$ (Theorem B), and the global sections $H^{0} (Ω, F)$ generate $F_{x}$ over $_{n} O_{x}$ for every $x \in Ω$ (Theorem A).

Proof. The argument runs in three movements: prove Theorem B for the structure sheaf from the $\overset{ˉ}{\partial}$ -estimates, bootstrap Theorem B to every coherent sheaf through local free resolutions, then deduce Theorem A from Theorem B.

Movement 1 — Theorem B for $_{n} O$ . Because $Ω$ is pseudoconvex 06.10.03, the Hörmander $L^{2}$ existence theorem 06.10.04 solves $\overset{ˉ}{\partial} u = f$ for every $\overset{ˉ}{\partial}$ -closed $(0, q)$ -form $f$ with $q \geq 1$ , in suitable weighted $L^{2}$ spaces, on $Ω$ . The Dolbeault isomorphism identifies $H^{q} (Ω,_{n} O)$ with the $q$ -th cohomology of the Dolbeault complex $(A^{0, ∙} (Ω), \overset{ˉ}{\partial})$ . Solvability of $\overset{ˉ}{\partial}$ in every positive bidegree is exactly the statement that this Dolbeault cohomology vanishes for $q \geq 1$ . Hence $H^{q} (Ω,_{n} O) = 0$ for $q \geq 1$ . The same estimate applied to vector-valued forms gives $H^{q} (Ω,_{n} O^{p}) = 0$ for every finite rank $p$ .

Movement 2 — dévissage to coherent $F$ . Fix a coherent $F$ . Oka's theorem 06.10.20 gives, locally near each point, a finite free resolution. Globalising on the Stein $Ω$ , a syzygy argument produces an exact sequence of coherent sheaves $$ 0 \to \mathcal{R} \to {}_n\mathcal{O}^{p_0} \xrightarrow{;\varphi;} \mathcal{F} \to 0, $$ where $R = ker φ$ is again coherent. The long exact cohomology sequence reads $$ \cdots \to H^q(\Omega, {}_n\mathcal{O}^{p_0}) \to H^q(\Omega, \mathcal{F}) \to H^{q+1}(\Omega, \mathcal{R}) \to H^{q+1}(\Omega, {}_n\mathcal{O}^{p_0}) \to \cdots. $$ For $q \geq 1$ the two outer $_{n} O^{p_{0}}$ -terms vanish by Movement 1, giving an isomorphism $H^{q} (Ω, F) ≅ H^{q + 1} (Ω, R)$ . Iterating the resolution one step further replaces $R$ by the next syzygy and shifts the cohomological degree up by one again. Since $Ω \subseteq C^{n}$ has cohomological dimension $n$ for coherent sheaves — Čech cohomology on a system of polydisc covers vanishes above degree $n$ — after at most $n$ shifts the target lands in a degree exceeding $n$ , where it is automatically zero. Tracing the chain of isomorphisms back gives $H^{q} (Ω, F) = 0$ for all $q \geq 1$ . This is Theorem B.

Movement 3 — Theorem A from Theorem B. Fix $x \in Ω$ and let $m_{x} \subset_{n} O_{x}$ be the maximal ideal. Consider the coherent subsheaf $F^{'} = m_{x} F$ obtained by multiplying $F$ by the ideal sheaf of the point $x$ , giving the exact sequence $$ 0 \to \mathcal{F}' \to \mathcal{F} \to \mathcal{F}/\mathcal{F}' \to 0, $$ whose quotient $F / F^{'}$ is a skyscraper at $x$ with stalk $F_{x} / m_{x} F_{x}$ . The long exact sequence in cohomology begins $$ H^0(\Omega, \mathcal{F}) \to H^0(\Omega, \mathcal{F}/\mathcal{F}') \to H^1(\Omega, \mathcal{F}'). $$ Theorem B applied to the coherent sheaf $F^{'}$ forces $H^{1} (Ω, F^{'}) = 0$ , so the first map is surjective. Its target is exactly the stalk $F_{x} / m_{x} F_{x}$ , and surjectivity there is, by Nakayama's lemma over the local ring $_{n} O_{x}$ , equivalent to the global sections generating the full stalk $F_{x}$ . This is Theorem A. $□$

Bridge. The dévissage in Movement 2 builds toward 06.10.22, where the same resolution-and-shift bootstrap is run on a complex space rather than an open subset of $C^{n}$ , and the structure-sheaf base case is supplied there by the finiteness theorems of Cartan–Serre. The foundational reason Theorem A follows from Theorem B is that the obstruction to lifting a stalk-level section to a global one lives in $H^{1}$ of a coherent kernel sheaf, and Theorem B annihilates that group; this is exactly the mechanism by which the additive Cousin problem becomes solvable, since its obstruction class is an element of $H^{1} (Ω,_{n} O)$ . The $\overset{ˉ}{\partial}$ -vanishing of 06.10.04 is dual to the holomorphic-convexity characterisation of Stein domains from 06.10.01, and putting these together identifies pseudoconvexity with the simultaneous vanishing of all coherent higher cohomology. The same vanishing appears again in the sheaf-theoretic route to the Levi problem 06.10.05, where global holomorphic functions separating points are produced by Theorem A. The central insight is that one analytic estimate — solvability of $\overset{ˉ}{\partial}$ — propagates through homological algebra to control every coherent sheaf at once, and the bridge is the local free resolution that turns an analytic input into a cohomological output.

Exercises Intermediate+

Exercise 3 (medium, short-answer).

Derive the solvability of the additive (Cousin I) problem on a Stein $Ω$ from Theorem B applied to $_{n} O$ .

Hint

The obstruction to gluing local meromorphic data with prescribed poles is a class in $H^{1} (Ω,_{n} O)$ .

Answer

Cousin I data is a collection of meromorphic functions $f_{i}$ on an open cover ${U_{i}}$ with $f_{i} - f_{j}$ holomorphic on $U_{i} \cap U_{j}$ . The differences $(f_{i} - f_{j})$ form a Čech $1$ -cocycle with values in $_{n} O$ , whose class in $H^{1} (Ω,_{n} O)$ is the obstruction to finding a global meromorphic $f$ with $f - f_{i}$ holomorphic on each $U_{i}$ . By Theorem B, $H^{1} (Ω,_{n} O) = 0$ , so the cocycle is a coboundary $f_{i} - f_{j} = h_{i} - h_{j}$ for holomorphic $h_{i}$ ; then $f := f_{i} - h_{i}$ is the desired global solution. Rubric: full credit for naming the $H^{1} (_{n} O)$ obstruction and the coboundary step.

Exercise 4 (medium, short-answer).

Use the exponential sheaf sequence $0 \to Z \to_{n} O e x p (2 π i \cdot)_{n} O^{*} \to 0$ and Theorem B to reduce the multiplicative (Cousin II) problem to a topological condition.

Hint

The obstruction lives in $H^{2} (Ω, Z)$ after Theorem B kills the analytic terms.

Answer

The long exact sequence in cohomology gives $H^{1} (Ω,_{n} O^{*}) \to H^{2} (Ω, Z) \to H^{2} (Ω,_{n} O)$ . Theorem B makes $H^{1} (Ω,_{n} O) = 0$ and $H^{2} (Ω,_{n} O) = 0$ , so the connecting maps yield $H^{1} (Ω,_{n} O^{*}) ≅ H^{2} (Ω, Z)$ . A Cousin II datum defines a class in $H^{1} (Ω,_{n} O^{*})$ ; it is solvable by a global holomorphic function with the prescribed divisor exactly when its image (the first Chern class) in $H^{2} (Ω, Z)$ vanishes. On a contractible Stein $Ω$ (e.g. a polydisc), $H^{2} (Ω, Z) = 0$ , so Cousin II is always solvable — Oka's principle. Rubric: full credit for the isomorphism $H^{1} (_{n} O^{*}) ≅ H^{2} (Ω, Z)$ and the topological criterion.

Exercise 6 (hard, short-answer).

Prove that on a Stein $Ω$ , every coherent ideal sheaf $I \subseteq_{n} O$ is globally generated, and deduce that the analytic subset $V (I)$ is the common zero locus of finitely many global holomorphic functions.

Hint

Apply Theorem A to $I$ at each point, then use coherence to pass from pointwise generation to a finite global generating set on a compact exhaustion.

Answer

By Theorem A applied to the coherent sheaf $I$ , the global sections $H^{0} (Ω, I)$ generate the stalk $I_{x}$ for every $x$ . Coherence means each stalk is finitely generated; by upper semicontinuity of the generation locus, a fixed finite set of global sections $g_{1}, \dots, g_{m} \in H^{0} (Ω, I)$ generates $I_{x}$ for all $x$ in a neighbourhood of any compact $K$ . Exhausting $Ω$ by compacts and using that $O (Ω)$ is a Fréchet algebra in which finitely generated submodules are closed, a countable—hence by the Fréchet structure a finite at each stalk—generating family suffices, and the common zero set of the $g_{j}$ is exactly $V (I)$ . Thus every analytic subset of a Stein domain cut out by a coherent ideal is the zero locus of global holomorphic functions. Rubric: full credit for the Theorem-A generation step and the passage to a global zero-set description.

Exercise 7 (hard, short-answer).

Show that Theorem B implies the Oka principle statement: on a Stein $Ω$ , a holomorphic line bundle is holomorphically isomorphic to the product bundle if and only if it is topologically so.

Hint

Compare $H^{1} (Ω,_{n} O^{*})$ with $H^{2} (Ω, Z)$ via the exponential sequence, then with the topological classification.

Answer

Holomorphic line bundles on $Ω$ are classified by $H^{1} (Ω,_{n} O^{*})$ ; topological complex line bundles by $H^{2} (Ω, Z)$ via the first Chern class. The exponential sequence and Theorem B (Exercise 4) give $H^{1} (Ω,_{n} O^{*}) c_{1} H^{2} (Ω, Z)$ an isomorphism, because $H^{1} (Ω,_{n} O) = H^{2} (Ω,_{n} O) = 0$ . Hence a holomorphic line bundle is holomorphically isomorphic to the product bundle (zero in $H^{1} (Ω,_{n} O^{*})$ ) precisely when its Chern class vanishes (zero in $H^{2} (Ω, Z)$ ), which is topological isomorphism to the product bundle. This is the line-bundle case of Grauert's Oka principle: on Stein spaces the holomorphic and topological classifications of bundles agree. Rubric: full credit for the exponential-sequence isomorphism and the holomorphic-equals-topological conclusion.

Lean formalization Intermediate+

Mathlib does not currently formalise coherent analytic sheaves on open subsets of $C^{n}$ , Stein open sets, or the cohomology $H^{q} (Ω, F)$ of a coherent analytic sheaf. A proposed signature, in Lean 4 / Mathlib syntax, sketching the target statements:

-- Sketch only; no current Mathlib coverage. See lean_mathlib_gap.
import Mathlib.Analysis.Analytic.Basic
import Mathlib.AlgebraicGeometry.Sheaves

namespace Codex.RiemannSurfaces.SeveralComplexVariables

variable {n : ℕ} (Ω : Set (Fin n → ℂ)) (hΩ : IsStein Ω)
variable (F : CoherentAnalyticSheaf Ω)

-- Theorem B: higher cohomology of a coherent sheaf on a Stein open vanishes.
theorem cartan_theorem_B (q : ℕ) (hq : 1 ≤ q) :
    SheafCohomology q Ω F.toSheaf = 0 := by
  sorry

-- Theorem A: global sections generate every stalk.
theorem cartan_theorem_A (x : Ω) :
    Submodule.span (StalkRing Ω x)
      (Set.range (fun s : F.globalSections => s.germ x)) = ⊤ := by
  sorry

end Codex.RiemannSurfaces.SeveralComplexVariables

The proof depends on names that do not exist in Mathlib: the coherent analytic sheaf type, the Stein predicate, sheaf cohomology of an analytic sheaf, and — load-bearing for Movement 1 — the Hörmander $\overset{ˉ}{\partial}$ -solvability theorem on pseudoconvex domains. Each is a candidate Mathlib contribution; until then this unit ships with lean_status: none.

Advanced results Master

Cartan's theorems are the foundation of the whole Stein theory, and several deeper statements fall out once they are in hand.

The Cartan–Serre finiteness theorem. On a compact complex manifold $X$ and coherent $F$ , the groups $H^{q} (X, F)$ are finite-dimensional over $C$ (Cartan–Serre 1953). The proof uses Theorem B in reverse: cover $X$ by Stein opens on which all higher cohomology vanishes, so the Čech complex computing $H^{q} (X, F)$ is built from Fréchet spaces of holomorphic sections; the Montel property of these spaces forces the cohomology to be finite-dimensional via L. Schwartz's theorem on compact perturbations of surjections between Fréchet spaces. Steinness of the cover is what makes the local terms tractable; compactness of $X$ is what makes the cover finite.

Solution of the Cousin problems. Cousin I (additive: prescribe principal parts) is solvable on every Stein $Ω$ because its obstruction lies in $H^{1} (Ω,_{n} O)$ , which Theorem B annihilates. Cousin II (multiplicative: prescribe divisors) is governed by $H^{1} (Ω,_{n} O^{*})$ ; the exponential sequence and Theorem B identify this with $H^{2} (Ω, Z)$ , so Cousin II is solvable precisely when the prescribed divisor's Chern class vanishes. On a contractible Stein domain this is automatic — the holomorphic problem reduces entirely to a topological one, the line-bundle case of Oka's principle.

Grauert's Oka principle. Grauert (1957–58) proved that on a Stein space the classification of holomorphic fibre bundles with complex Lie structure group agrees with the topological classification: the inclusion of holomorphic into continuous sections is a weak homotopy equivalence for such bundles. Theorems A and B are the abelian-sheaf core of this non-abelian statement; Grauert's contribution was to push the vanishing from coherent sheaves to non-abelian $H^{1}$ with values in a sheaf of holomorphic maps to a Lie group.

Embedding and the converse. A connected $n$ -dimensional Stein manifold embeds properly and holomorphically into $C^{2 n + 1}$ (Remmert; Bishop–Narasimhan), and Theorem A is what produces enough global holomorphic functions to separate points and give local coordinates. Conversely, a complex manifold on which Theorem B holds for all coherent sheaves and which is holomorphically separable is Stein — so Theorems A and B do not merely hold on Stein manifolds, their conjunction with separability characterises Steinness.

Higher dévissage and the cohomological dimension. The bootstrap in Movement 2 of the key proof exploits that an open subset of $C^{n}$ has coherent-cohomological dimension at most $n$ . On a general Stein space of dimension $d$ the same bound holds with $n$ replaced by $d$ , and the dévissage still terminates; this is the form needed in 06.10.22.

Synthesis. Cartan's two theorems are the foundational reason the geometry of Stein domains is governed entirely by global holomorphic functions: Theorem A makes the global section module rich enough to generate every local germ, and Theorem B removes every cohomological obstruction to assembling local data globally. This is exactly the mechanism that solves the Cousin problems — the additive problem because its obstruction in $H^{1} (Ω,_{n} O)$ vanishes, the multiplicative problem because the residual obstruction is purely topological in $H^{2} (Ω, Z)$ . The vanishing of higher coherent cohomology is dual to the holomorphic-convexity characterisation of Steinness, so the analytic input — solvability of $\overset{ˉ}{\partial}$ from the Levi problem 06.10.04, 06.10.05 — and the homological output are two readings of one fact. Putting these together, the dévissage that propagates structure-sheaf vanishing to every coherent sheaf identifies the entire coherent-cohomological behaviour of a Stein domain with a single $\overset{ˉ}{\partial}$ -estimate; this is the central insight that organises the Stein theory and the bridge to the finiteness theorems on compact manifolds and the coherence theory on complex spaces.

Full proof set Master

Proposition 1 (Theorem B for the structure sheaf via $\overset{ˉ}{\partial}$ ). Let $Ω \subseteq C^{n}$ be pseudoconvex. Then $H^{q} (Ω,_{n} O) = 0$ for every $q \geq 1$ .

Proof. The Dolbeault resolution $0 \to_{n} O \to A^{0, 0} \overset{ˉ}{\partial} A^{0, 1} \overset{ˉ}{\partial} \dots$ by sheaves of smooth $(0, ∙)$ -forms is a resolution of $_{n} O$ by fine (hence acyclic) sheaves, so $H^{q} (Ω,_{n} O)$ equals the $q$ -th cohomology of the complex of global forms $(A^{0, ∙} (Ω), \overset{ˉ}{\partial})$ . For $q \geq 1$ , given a $\overset{ˉ}{\partial}$ -closed $(0, q)$ -form $f$ , the Hörmander $L^{2}$ theorem 06.10.04 on the pseudoconvex $Ω$ produces $u$ with $\overset{ˉ}{\partial} u = f$ ; elliptic regularity of $\overset{ˉ}{\partial}$ in the interior promotes the $L^{2}$ solution to a smooth one. Thus every closed form in positive bidegree is exact, and the Dolbeault cohomology vanishes in those degrees. The vector-valued case $_{n} O^{p}$ follows componentwise. $□$

Proposition 2 (dévissage). Let $Ω$ be Stein and $F$ coherent. If $H^{q} (Ω, G) = 0$ for all $q \geq 1$ and all coherent $G$ that are quotients of finite free sheaves appearing in a resolution of $F$ , then $H^{q} (Ω, F) = 0$ for all $q \geq 1$ .

Proof. Choose a presentation $0 \to R \to_{n} O^{p_{0}} \to F \to 0$ with $R$ coherent (Oka 06.10.20). The long exact sequence and Proposition 1 give $H^{q} (Ω, F) ≅ H^{q + 1} (Ω, R)$ for $q \geq 1$ . Replacing $F$ by $R$ and iterating produces $H^{q} (Ω, F) ≅ H^{q + k} (Ω, R_{k})$ for the $k$ -th syzygy $R_{k}$ . Since coherent Čech cohomology on $Ω \subseteq C^{n}$ vanishes above degree $n$ (a polydisc cover has no non-degenerate $(n + 1)$ -fold intersections contributing), choosing $k > n$ forces the right side to zero. $□$

Proposition 3 (Nakayama step for Theorem A). Let $Ω$ be Stein, $F$ coherent, and suppose $H^{1} (Ω, m_{x} F) = 0$ for a point $x$ . Then $H^{0} (Ω, F)$ generates $F_{x}$ over $_{n} O_{x}$ .

Proof. The skyscraper sequence $0 \to m_{x} F \to F \to F / m_{x} F \to 0$ has $(F / m_{x} F)$ supported at $x$ with stalk the finite-dimensional vector space $F_{x} / m_{x} F_{x}$ . The cohomology sequence gives surjectivity of $H^{0} (Ω, F) \to F_{x} / m_{x} F_{x}$ once $H^{1} (Ω, m_{x} F) = 0$ . Lift a $C$ -basis of $F_{x} / m_{x} F_{x}$ to global sections $s_{1}, \dots, s_{r}$ ; their germs generate $F_{x}$ modulo $m_{x} F_{x}$ . Nakayama's lemma over the local ring $_{n} O_{x}$ (with $F_{x}$ finitely generated by coherence) upgrades this to generation of $F_{x}$ itself. $□$

Corollary (Cousin I). On a Stein $Ω$ , every additive Cousin datum is solvable.

Proof. The obstruction is the image of the datum under $\overset{ˇ}{H}^{0} ({U_{i}}, M /_{n} O) \to \overset{ˇ}{H}^{1} ({U_{i}},_{n} O)$ , and $H^{1} (Ω,_{n} O) = 0$ by Proposition 1, so the obstruction vanishes and a global solution exists. $□$

Corollary (Cousin II / Oka). On a Stein $Ω$ with $H^{2} (Ω, Z) = 0$ , every multiplicative Cousin datum is solvable.

Proof. The exponential sequence gives $H^{1} (Ω,_{n} O^{*}) ≅ H^{2} (Ω, Z)$ after Proposition 1 kills the surrounding $_{n} O$ -terms. When $H^{2} (Ω, Z) = 0$ the multiplicative obstruction vanishes, so the prescribed divisor is the divisor of a global holomorphic function. $□$

Connections Master

Coherent analytic sheaves and Oka's coherence theorem 06.10.20. This is the structural input on which both theorems rest: Oka's coherence of $_{n} O$ is what makes the category of coherent analytic sheaves stable under the kernels and finite presentations used in the dévissage. Without it the syzygy sheaves $R_{k}$ in Movement 2 might fail to be coherent and the bootstrap would not close.
The $\overset{ˉ}{\partial}$ -equation and Hörmander $L^{2}$ estimates 06.10.04. The analytic engine of Theorem B for the structure sheaf. Solvability of $\overset{ˉ}{\partial}$ on a pseudoconvex domain is, through the Dolbeault isomorphism, identical to the vanishing $H^{q} (Ω,_{n} O) = 0$ that seeds the entire proof.
Solution of the Levi problem 06.10.05. This unit certifies that pseudoconvex equals domain-of-holomorphy equals Stein, the hypothesis under which Cartan's theorems hold; the sheaf-theoretic converse — using Theorem A to build separating global functions — is one route back into the Levi problem.
Cousin I/II and the Levi problem in $C^{n}$ 06.10.11. The Cousin problems are the headline corollaries of Theorem B; that unit develops the classical formulation, while the present unit supplies the cohomological vanishing that explains why the additive problem is always solvable and the multiplicative one reduces to topology.
Complex spaces and coherence on them 06.10.22. The forward generalisation: Theorems A and B are restated for coherent sheaves on Stein complex spaces, where the structure-sheaf base case is supplied by the Cartan–Serre finiteness machinery rather than by a flat $\overset{ˉ}{\partial}$ -estimate, and the dévissage runs with cohomological dimension equal to the space's dimension.
Domains of holomorphy and holomorphic convexity 06.10.01. The vanishing of all coherent higher cohomology is dual to holomorphic convexity; this unit's theorems are the cohomological face of the same Steinness that 06.10.01 describes geometrically.

Historical & philosophical context Master

Henri Cartan formulated and proved Theorems A and B in his Séminaire at the École Normale Supérieure across 1951–1953 ^{[Cartan 1951]}, building on Kiyoshi Oka's coherence theorems and on the notion of faisceau (sheaf) that Jean Leray had introduced in the 1940s. The labels "A" and "B" are Cartan's own, attached to the two consecutive exposés in which the global-generation and cohomology-vanishing statements were established for what he called variétés de Stein, after Karl Stein's 1951 characterisation of the manifolds on which the Cousin and Poincaré problems are uniformly solvable.

The cohomological vanishing was made possible by Oka's 1950 coherence theorem for the sheaf of holomorphic functions; Cartan recognised that Oka's local statements about ideals of holomorphic functions were exactly the coherence needed to run a sheaf-theoretic dévissage. The companion finiteness theorem for compact complex manifolds was announced jointly by Cartan and Jean-Pierre Serre in a 1953 Comptes Rendus note ^{[Cartan-Serre 1953]}, using Theorem B on a Stein cover together with L. Schwartz's functional-analytic argument.

Serre's 1956 paper Géométrie algébrique et géométrie analytique (the GAGA paper) transported this analytic framework into algebraic geometry, where the coherent-sheaf cohomology of projective varieties matches its analytic counterpart; the Stein condition is the analytic analogue of affineness, and Cartan's Theorem B is the analytic shadow of the vanishing of higher cohomology of quasi-coherent sheaves on affine schemes. Hörmander's 1965 $L^{2}$ method ^{[Hörmander 1990]} gave the proof its modern analytic backbone, replacing the original sheaf-cohomology bookkeeping for the structure sheaf with a single weighted estimate for $\overset{ˉ}{\partial}$ , the form in which the theorems are most often proved today.

Bibliography Master

@misc{Cartan1951Seminaire,
  author       = {Cartan, Henri},
  title        = {S{\'e}minaire E.N.S. 1951--1952: Fonctions analytiques de plusieurs variables complexes},
  year         = {1952},
  howpublished = {{\'E}cole Normale Sup{\'e}rieure, Paris},
  note         = {Expos{\'e}s containing Theorems A and B for coherent analytic sheaves on Stein open sets}
}

@article{CartanSerre1953,
  author  = {Cartan, Henri and Serre, Jean-Pierre},
  title   = {Un th{\'e}or{\`e}me de finitude concernant les vari{\'e}t{\'e}s analytiques compactes},
  journal = {C. R. Acad. Sci. Paris},
  volume  = {237},
  year    = {1953},
  pages   = {128--130}
}

@article{Oka1950Coherence,
  author  = {Oka, Kiyoshi},
  title   = {Sur les fonctions analytiques de plusieurs variables. VII. Sur quelques notions arithm{\'e}tiques},
  journal = {Bull. Soc. Math. France},
  volume  = {78},
  year    = {1950},
  pages   = {1--27}
}

@article{Serre1956GAGA,
  author  = {Serre, Jean-Pierre},
  title   = {G{\'e}om{\'e}trie alg{\'e}brique et g{\'e}om{\'e}trie analytique},
  journal = {Ann. Inst. Fourier (Grenoble)},
  volume  = {6},
  year    = {1956},
  pages   = {1--42}
}

@book{Hormander1990SCV,
  author    = {H{\"o}rmander, Lars},
  title     = {An Introduction to Complex Analysis in Several Variables},
  edition   = {3rd},
  publisher = {North-Holland},
  year      = {1990}
}

@book{GunningRossi1965,
  author    = {Gunning, Robert C. and Rossi, Hugo},
  title     = {Analytic Functions of Several Complex Variables},
  publisher = {Prentice-Hall},
  year      = {1965}
}

@book{GunningVolIII,
  author    = {Gunning, Robert C.},
  title     = {Introduction to Holomorphic Functions of Several Variables, Vol. III: Homological Theory},
  publisher = {Wadsworth \& Brooks/Cole},
  year      = {1990}
}

@book{GrauertRemmertStein,
  author    = {Grauert, Hans and Remmert, Reinhold},
  title     = {Theory of Stein Spaces},
  series    = {Grundlehren der mathematischen Wissenschaften},
  volume    = {236},
  publisher = {Springer},
  year      = {1979}
}

@article{Grauert1958Oka,
  author  = {Grauert, Hans},
  title   = {Analytische Faserungen {\"u}ber holomorph-vollst{\"a}ndigen R{\"a}umen},
  journal = {Math. Ann.},
  volume  = {135},
  year    = {1958},
  pages   = {263--273}
}

Prerequisites

06.10.20
06.10.04
06.10.05
06.10.11

Tier anchors

beginner: Krantz *Function Theory of Several Complex Variables* §7 (informal Cousin discussion); Gunning–Rossi Ch. VIII opening
intermediate: Hörmander *An Introduction to Complex Analysis in Several Variables* §7.2–§7.4; Gunning–Rossi Ch. VIII §A–§C
master: Cartan, Séminaire ENS 1951–53 (originator); Gunning *Introduction to Holomorphic Functions of Several Variables, Vol. III: Homological Theory* Ch. on Theorems A and B; Grauert–Remmert *Theory of Stein Spaces* §IV–§V

References

Cartan — Séminaire E.N.S. 1951–52, Fonctions analytiques de plusieurs variables complexes · Exposés XVIII–XX (the cohomological vanishing theorems for coherent analytic sheaves on Stein open sets; Theorems A and B)
Cartan–Serre — Un théorème de finitude concernant les variétés analytiques compactes · C. R. Acad. Sci. Paris 237 (1953), 128–130 (finiteness and the coherent-sheaf vanishing framework)
Hörmander — An Introduction to Complex Analysis in Several Variables · North-Holland 1990 (3rd ed.), §7.2–§7.4 (the $\bar\partial$ route to Theorem B and the Cousin problems)
Gunning — Introduction to Holomorphic Functions of Several Variables, Vol. III: Homological Theory · Wadsworth & Brooks/Cole 1990, chapters on coherent analytic sheaves and Theorems A and B
Gunning–Rossi — Analytic Functions of Several Complex Variables · Prentice-Hall 1965, Ch. VIII (Theorems A and B for Stein manifolds; the Cousin and Oka applications)
Grauert–Remmert — Theory of Stein Spaces · Springer Grundlehren 236, 1979, §IV–§V (Theorems A and B; Oka's coherence; the Cousin and Poincaré problems)
Krantz — Function Theory of Several Complex Variables · AMS Chelsea 2001 (2nd ed.), §7 (Cousin problems and the additive/multiplicative obstructions)

Estimated time

beginner: 16m
intermediate: 42m
master: 70m