06.09.01 · riemann-surfaces / stein

Stein Riemann surfaces

shipped3 tiersLean: none

Anchor (Master): Behnke-Stein 1949 *Entwicklung analytischer Funktionen auf Riemannschen Flächen* (originator: every non-compact RS is Stein); Stein 1951 *Analytische Funktionen mehrerer komplexer Veränderlichen zu vorgegebenen Periodizitätsmoduln und das zweite Cousinsche Problem* (originator: Stein-manifold definition); Cartan 1951–53 séminaire (Theorems A and B); Forster *Lectures on Riemann Surfaces* §25–§26; Hörmander *Several Complex Variables* Ch. IV–VI; Cieliebak-Eliashberg *From Stein to Weinstein and Back* (2012)

Intuition [Beginner]

Take a non-compact Riemann surface $X$ — a one-complex-dimensional analytic space without the closed-and-bounded property of a compact curve. Examples are the complex line $C$ , the punctured line $C^{*}$ , the open unit disc, an annulus, the upper half-plane, the universal cover of any genus- $g \geq 2$ surface, and any compact Riemann surface with finitely many points removed. The driving question is: how many global holomorphic functions does $X$ carry, and are there enough of them to do useful global geometry?

A non-compact $X$ is called Stein when two simple conditions hold. First, the global holomorphic functions on $X$ are abundant enough to separate points — given two distinct points, some holomorphic function takes different values on them. Second, the global holomorphic functions are holomorphically convex — the hull of a compact set, formed by adding every point that obeys the same maximum-modulus inequalities the compact set does, is itself compact and does not balloon out to infinity.

The remarkable theorem of Behnke and Stein from 1949 says: every non-compact Riemann surface is Stein. There is no non-Stein non-compact Riemann surface. The compact-versus-non-compact dichotomy on a Riemann surface is exactly the dichotomy between curves carrying very few global holomorphic functions (constants only, in the compact case) and curves carrying many global holomorphic functions (a Stein wealth, in the non-compact case). In higher complex dimension this dichotomy is no longer automatic — non-compact non-Stein manifolds exist — and Stein theory becomes a substantive classification question. On a Riemann surface, the answer to a deep question takes its simplest possible form.

Visual [Beginner]

A schematic of a non-compact Riemann surface $X$ shown as the open complex line, with a compact disc $K$ shaded inside it and the holomorphic hull $\hat{K}$ drawn as a slightly larger but still compact disc enclosing $K$ . A second panel shows two distinct points $p, q$ on $X$ with a holomorphic function $f$ depicted as a level-set surface that takes different heights at $p$ and $q$ , illustrating point-separation. A third panel shows a strictly subharmonic exhaustion $ϕ : X \to R$ as a real-valued landscape over $X$ , with level sets compact and rising to infinity at the ideal boundary.

Worked example [Beginner]

Take $X = C$ , the complex line. Pick a compact set $K = \overline{D}_{r}$ , the closed disc of radius $r$ around the origin. Compute its holomorphic hull.

For each holomorphic function $f \in O (C)$ , the maximum modulus on $K$ is $sup_{∣ z ∣ \leq r} ∣ f (z) ∣ = M_{f}$ . The hull condition says $\hat{K} = {z \in C : ∣ f (z) ∣ \leq M_{f} for every entire f}$ . Take $f (z) = z^{n}$ : the condition reads $∣ z ∣^{n} \leq r^{n}$ , equivalently $∣ z ∣ \leq r$ . The hull is contained in the disc of radius $r$ , and the disc itself is contained in the hull (every point in $K$ satisfies the inequality on $K$ ). The hull equals $K$ .

Repeat for any compact $K \subset C$ : by an analogous argument with monomials and Runge approximation, the hull $\hat{K}$ is contained in the convex hull of $K$ (a bounded region) — compact, in particular. Holomorphic convexity holds.

Point separation on $C$ : pick $p, q$ distinct. The function $f (z) = z$ takes values $p$ and $q$ , distinct. Done.

What this tells us: $C$ satisfies both Stein conditions, and is the simplest example of a Stein Riemann surface. The Behnke-Stein theorem says every non-compact Riemann surface behaves the same way — point-separation and holomorphic convexity are automatic on a non-compact Riemann surface.

Check your understanding [Beginner]

Exercise (easy, multiple choice).

A Riemann surface $X$ is holomorphically convex when:

A. Every holomorphic function on $X$ is bounded. B. For every compact $K \subset X$ , the holomorphic hull $\hat{K}$ of $K$ is compact. C. The space $X$ has a convex open cover. D. Every two points of $X$ are separated by a holomorphic function.

Hint

The hull is built from inequalities of the form $∣ f (z) ∣ \leq sup_{K} ∣ f ∣$ over all global holomorphic $f$ .

Answer

B. Feedback-correct: holomorphic convexity is the compactness of $\hat{K}$ for every compact $K$ , where $\hat{K}$ is the set of points obeying the maximum-modulus inequalities defined by $K$ and the global holomorphic functions. Feedback-wrong: A is false on every non-compact Stein surface (entire functions are typically unbounded); C confuses topological convexity with the holomorphic version; D describes point separation, the other Stein axiom.

Formal definition [Intermediate+]

Let $X$ be a connected non-compact Riemann surface and $O (X)$ its $C$ -algebra of global holomorphic functions.

Definition. $X$ is holomorphically convex when for every compact subset $K \subset X$ the holomorphic hull

\hat{K} = {z \in X : ∣ f (z) ∣ \leq w \in K sup ∣ f (w) ∣ for every f \in O (X)}

is compact in $X$ .

Definition. $X$ is holomorphically separable when for every pair of distinct points $p \neq = q$ in $X$ there exists $f \in O (X)$ with $f (p) \neq = f (q)$ .

Definition (Stein 1951). A non-compact Riemann surface $X$ is a Stein surface if it is holomorphically convex and holomorphically separable.

Theorem (Behnke-Stein 1949). Every connected non-compact Riemann surface is Stein.

Equivalent forms. Let $X$ be a connected non-compact Riemann surface. The following are equivalent.

$X$ is Stein (i.e., holomorphically convex and holomorphically separable).
There exists a strictly subharmonic exhaustion $ϕ : X \to R$ , meaning $ϕ$ is $C^{\infty}$ , $ϕ^{- 1} ((- \infty, c])$ is compact for every $c \in R$ , and the Levi form $\partial \overset{ˉ}{\partial} ϕ$ is everywhere positive.
$H^{q} (X, F) = 0$ for every coherent analytic sheaf $F$ on $X$ and every $q \geq 1$ (Theorem B).
The $\overset{ˉ}{\partial}$ -equation $\overset{ˉ}{\partial} f = g$ is solvable for every $g \in C_{c}^{\infty} (X)$ with no compatibility condition (this restates Theorem B for $F = O_{X}$ and $q = 1$ via the Dolbeault comparison 06.04.02).

Notation. $O (X) = Γ (X, O_{X})$ denotes global holomorphic functions; $M (X) = Γ (X, M_{X})$ denotes global meromorphic functions; $O^{*} (X)$ denotes the units of $O (X)$ .

Counterexamples to common slips

A compact Riemann surface satisfies neither Stein axiom: $O (X) = C$ for compact connected $X$ , so points cannot be separated and the maximum modulus principle obstructs the holomorphic-convexity formula. The Stein definition explicitly requires non-compactness.
Holomorphic convexity is not topological convexity. The complex line $C$ is holomorphically convex; so is the punctured line $C^{*}$ , even though it is not topologically convex. The hull is built from holomorphic inequalities, not affine line segments.
The compactness of $\hat{K}$ must hold for every compact $K$ , not just for one. A surface where some hulls are compact and others are not (impossible for Riemann surfaces but possible in higher-dimensional analytic spaces) is not Stein.
Point separation alone is not enough in higher dimension: there exist non-Stein complex manifolds where holomorphic functions separate points but the hull condition fails (Eichler's example in dimension $\geq 2$ ). On a Riemann surface this distinction is moot — both conditions are automatic on non-compact $X$ — but the formal definition keeps both axioms because the higher-dimensional theory needs them separately.

Key theorem with proof [Intermediate+]

Theorem (Behnke-Stein 1949). Every connected non-compact Riemann surface $X$ is Stein.

Proof. The argument has three steps: construct a strictly subharmonic exhaustion via a Runge-type approximation, derive holomorphic convexity from the exhaustion, and derive holomorphic separability from $\overset{ˉ}{\partial}$ -solvability and Mittag-Leffler interpolation.

Step 1 — strictly subharmonic exhaustion. The space $X$ is non-compact, connected, paracompact, and second-countable (a Riemann surface is automatically second-countable by Radó's theorem). Choose an exhaustion of $X$ by relatively compact open sets $X_{1} ⋐ X_{2} ⋐ \dots$ with $⋃_{n} X_{n} = X$ and each $X_{n}$ Runge in $X$ (every $f \in O (X_{n})$ is uniformly approximable on every compact $K \subset X_{n}$ by functions in $O (X)$ ). Such an exhaustion exists by the Behnke-Stein construction: start with any countable cover by relatively compact open discs, take iterated Runge envelopes, and verify by compactness arguments that the resulting nested sequence is exhausting.

On each $X_{n}$ pick a smooth strictly subharmonic function $ϕ_{n} : X_{n} \to R$ with $ϕ_{n} \to + \infty$ on $\partial X_{n}$ — the standard model is $ϕ_{n} (z) = - lo g d (z, \partial X_{n})^{- 1} + ∣ z ∣^{2}$ in a local chart, glued by a partition of unity. By a uniform-approximation argument (Runge-type, using the Runge property of each $X_{n}$ ), the $ϕ_{n}$ can be chosen to agree on $X_{n - 1}$ up to a uniform error decaying with $n$ . The limit $ϕ = lim_{n} ϕ_{n}$ exists, is smooth, strictly subharmonic, and exhausting on $X$ .

Step 2 — holomorphic convexity from $ϕ$ . Let $K \subset X$ be compact. Pick $c > sup_{K} ϕ$ and set $K_{c} = ϕ^{- 1} ((- \infty, c])$ , which is compact since $ϕ$ is exhausting. The holomorphic hull $\hat{K}$ satisfies $\hat{K} \subset K_{c}$ : for any $z \in X ∖ K_{c}$ , $ϕ (z) > c \geq sup_{K} ϕ$ , and a standard argument (Hörmander's $L^{2}$ -method, or the elementary Forster argument via Runge approximation) constructs an entire $f \in O (X)$ with $∣ f (z) ∣ > sup_{K} ∣ f ∣$ , witnessing $z \in / \hat{K}$ . Specifically: solve the $\overset{ˉ}{\partial}$ -equation $\overset{ˉ}{\partial} u = \overset{ˉ}{\partial} (χ e^{N ϕ})$ on $X$ for $χ$ a cutoff supported in a small neighbourhood of $z$ ; the solution $u$ exists by Hörmander's $L^{2}$ -theorem in dimension one (or by Forster §28 directly), and $f = χ e^{N ϕ} - u$ is entire. For $N$ large, $∣ f (z) ∣$ exceeds $sup_{K} ∣ f ∣$ . So $\hat{K} \subset K_{c}$ , and $\hat{K}$ is closed in $X$ and contained in the compact $K_{c}$ , hence compact. Holomorphic convexity holds.

Step 3 — holomorphic separability. Let $p \neq = q$ be distinct points of $X$ . The Mittag-Leffler problem asks for a global meromorphic function with a prescribed simple pole at $p$ and zero at $q$ (and holomorphic elsewhere); this is the prototypical Cousin I problem on $X$ . The obstruction class lives in $H^{1} (X, O_{X})$ . By Step 2 plus the $\overset{ˉ}{\partial}$ -solvability $\overset{ˉ}{\partial} u = g$ on $X$ (Hörmander 1-d / Forster §28), $H^{1} (X, O_{X}) = 0$ on every non-compact Riemann surface. Therefore the Cousin I datum is realised by a global meromorphic $f \in M (X)$ , and the function $g (z) = 1/ f (z)$ when $f (p) \neq = 0$ , or a similar Mittag-Leffler construction at $p$ alone, produces a holomorphic $h \in O (X)$ with $h (p) \neq = h (q)$ . (More directly: a global solution to $\overset{ˉ}{\partial}$ -equations gives explicit holomorphic functions on $X$ with prescribed values at any finite set of points, by interpolation.) Holomorphic separability holds. $□$

The three-step structure follows Forster §25–§26 (Runge-exhaustion + $\overset{ˉ}{\partial}$ -solvability + Mittag-Leffler) and dovetails with Donaldson §11 (PSH-exhaustion + Hörmander's $L^{2}$ ). The two routes prove the same theorem; Forster's route is more elementary in dimension one, while Hörmander's $L^{2}$ -method generalises immediately to Stein manifolds in arbitrary complex dimension.

Bridge. The theorem proven here is the foundational input to the Cartan-Serre Theorems A and B for Stein Riemann surfaces (the planned successor unit 06.09.02): once $X$ is Stein, every coherent analytic sheaf on $X$ is generated by global sections (Theorem A) and has vanishing higher cohomology (Theorem B). Combined with the Čech construction 06.04.02, this means every cohomological obstruction on a non-compact Riemann surface — Mittag-Leffler, Cousin I, Cousin II, holomorphic line-bundle classification, the $\overset{ˉ}{\partial}$ -equation — collapses to a single statement: $H^{1} (X, F) = 0$ for every coherent $F$ . The compact case 06.04.01 sees the same Čech machinery produce the Riemann-Roch dimension count $dim H^{0} (L) - dim H^{1} (L) = de g L + 1 - g$ , with $H^{1}$ generically non-zero; the non-compact case sees $H^{1}$ disappear entirely. Putting these together, the foundational insight is that the Stein property on a Riemann surface is the automatic analytic enrichment of non-compactness, and the entire analytic-spine of complex geometry on a non-compact curve — function existence, line-bundle triviality, $\overset{ˉ}{\partial}$ -solvability — is a corollary of one structural fact: the existence of a strictly subharmonic exhaustion.

Exercises [Intermediate+]

Exercise 4 (medium, short-answer).

Show that every holomorphic line bundle $L$ on a non-compact connected Riemann surface $X$ is holomorphically isomorphic to the structure-sheaf bundle $O_{X}$ .

Hint

The Picard group is $Pic (X) = H^{1} (X, O_{X}^{*})$ . Use the exponential exact sequence and Theorem B on $O_{X}$ .

Answer

The exponential exact sequence of sheaves $0 \to Z \to O_{X} e x p (2 π i \cdot) O_{X}^{*} \to 0$ on $X$ produces a long exact sequence in cohomology

\dots \to H^{1} (X, Z) \to H^{1} (X, O_{X}) \to H^{1} (X, O_{X}^{*}) \to H^{2} (X, Z) \to \dots .

By Behnke-Stein and Theorem B, $H^{1} (X, O_{X}) = 0$ . By the topology of a non-compact connected real surface, $H^{2} (X, Z) = 0$ (a non-compact connected surface has no top-dimensional cohomology with $Z$ -coefficients). Therefore $H^{1} (X, O_{X}^{*}) = 0$ , i.e. $Pic (X) = 0$ , and every holomorphic line bundle is isomorphic to $O_{X}$ . Rubric: full credit for the exponential exact sequence + Theorem B + topological vanishing of $H^{2}$ .

Exercise 5 (medium, short-answer).

Show that the Mittag-Leffler problem on a non-compact Riemann surface $X$ is always solvable: given finitely many points $p_{1}, \dots, p_{k} \in X$ and finitely many principal-part data at each point (a finite Laurent tail in any local coordinate at $p_{i}$ ), there exists a global meromorphic function $f \in M (X)$ with the prescribed principal parts at each $p_{i}$ and holomorphic elsewhere.

Hint

The obstruction class lives in $H^{1} (X, O_{X})$ . Apply Behnke-Stein.

Answer

The principal-parts sheaf $M_{X} / O_{X}$ is the quotient of meromorphic by holomorphic sections; a Mittag-Leffler datum is a global section $μ \in H^{0} (X, M_{X} / O_{X})$ . The short exact sequence $0 \to O_{X} \to M_{X} \to M_{X} / O_{X} \to 0$ produces the long exact sequence

\dots \to H^{0} (X, M_{X}) \to H^{0} (X, M_{X} / O_{X}) δ H^{1} (X, O_{X}) \to \dots .

The obstruction to lifting $μ$ to a global meromorphic function is $δ (μ) \in H^{1} (X, O_{X})$ . By Behnke-Stein, $X$ is Stein and Theorem B gives $H^{1} (X, O_{X}) = 0$ . Hence every Mittag-Leffler datum lifts. Rubric: full credit for the principal-parts sheaf + connecting map + Theorem B vanishing.

Exercise 6 (medium, short-answer).

Show that on a non-compact Riemann surface $X$ , the Cousin II problem (prescribed local divisors realised by a single global meromorphic function with no extra zeros or poles) is always solvable.

Hint

Cousin II reduces to triviality of a class in $H^{1} (X, O_{X}^{*})$ followed by Cousin I.

Answer

A Cousin II datum is a covering ${U_{i}}$ of $X$ together with non-zero meromorphic functions $f_{i} \in M^{*} (U_{i})$ such that $f_{i} / f_{j} \in O^{*} (U_{i} \cap U_{j})$ on each overlap. The cocycle $(f_{i} / f_{j})$ defines a class in $H^{1} (X, O_{X}^{*}) = Pic (X)$ . By Exercise 4, $Pic (X) = 0$ , so the cocycle is a coboundary $f_{i} / f_{j} = h_{i} / h_{j}$ for some $h_{i} \in O^{*} (U_{i})$ . The global function $g_{i} = f_{i} / h_{i}$ then satisfies $g_{i} = g_{j}$ on overlaps and assembles into a global meromorphic $g \in M (X)$ with the prescribed local divisors. Rubric: full credit for the Pic-vanishing reduction and the patching argument.

Exercise 7 (hard, short-answer).

Let $X$ be a non-compact connected Riemann surface and let $K \subset X$ be compact such that $X ∖ K$ has no relatively compact connected components. Show that every $f \in O (U)$ holomorphic on a neighbourhood $U$ of $K$ is uniformly approximable on $K$ by functions in $O (X)$ (Runge approximation on a Stein Riemann surface).

Hint

Express $f$ via a Cauchy-type integral with kernel of compact support, solve the resulting $\overset{ˉ}{\partial}$ -problem on $X$ , and combine the local datum with the Stein vanishing.

Answer

Pick a smooth cutoff $χ$ with compact support in $U$ and $χ \equiv 1$ on a neighbourhood of $K$ . The function $χ f$ extends by zero to a $C^{\infty}$ function on $X$ , with $\overset{ˉ}{\partial} (χ f) = (\overset{ˉ}{\partial} χ) f$ supported in the compact annulus $U ∖ K$ . By Behnke-Stein and the $\overset{ˉ}{\partial}$ -solvability $\overset{ˉ}{\partial} u = g$ on the Stein surface $X$ , there exists $u \in C^{\infty} (X)$ with $\overset{ˉ}{\partial} u = (\overset{ˉ}{\partial} χ) f$ . The function $F = χ f - u$ is then holomorphic on all of $X$ (since $\overset{ˉ}{\partial} F = 0$ globally), and equals $f - u$ on a neighbourhood of $K$ . The error $u$ is small in $L^{2}$ -norm (or sup norm, after Sobolev embedding); the hypothesis " $X ∖ K$ has no relatively compact connected components" is what allows the local datum on $U$ to extend to one on a slightly larger Runge open without picking up homological obstructions from bounded components. Iterating with shrinking error gives uniform approximation on $K$ . Rubric: full credit for the cutoff + $\overset{ˉ}{\partial}$ -solve + global-holomorphy chain.

Exercise 8 (hard, short-answer).

Construct a strictly subharmonic exhaustion $ϕ : C \to R$ explicitly, and verify the three Stein-equivalent properties for $C$ from this single $ϕ$ .

Hint

Take $ϕ (z) = ∣ z ∣^{2}$ and check that $\partial \overset{ˉ}{\partial} ϕ$ is positive everywhere, $ϕ$ is exhausting, and the level sets $ϕ \leq c$ are compact.

Answer

Take $ϕ (z) = ∣ z ∣^{2} = z \overset{z}{ˉ}$ . Compute $\partial ϕ = \overset{z}{ˉ} d z$ , $\overset{ˉ}{\partial} ϕ = z d \overset{z}{ˉ}$ , so $\partial \overset{ˉ}{\partial} ϕ = - d z \land d \overset{z}{ˉ} = 2 i d x \land d y$ (the standard volume form, up to a constant), strictly positive as a $(1, 1)$ -form. The level set $ϕ \leq c$ is the closed disc of radius $c$ , compact. As $∣ z ∣ \to \infty$ , $ϕ (z) \to \infty$ , so $ϕ$ is exhausting. Three Stein-equivalent consequences:

(i) Holomorphic convexity: for any compact $K$ contained in $ϕ \leq c$ , $\hat{K} \subset ϕ \leq c$ is compact.

(ii) Theorem B: $H^{1} (C, O) = 0$ , equivalent to $\overset{ˉ}{\partial}$ -solvability — Hörmander's $L^{2}$ -method on a strictly plurisubharmonic exhaustion produces $u$ with $\overset{ˉ}{\partial} u = g$ for any compactly supported $g$ .

(iii) Holomorphic separability: $f (z) = z$ separates points immediately, but the deeper fact is that $\overset{ˉ}{\partial}$ -solvability constructs $f \in O (C)$ taking any prescribed finite jet at any finite set of points, which is much stronger than separation.

Rubric: full credit for the verification of strictly positive Levi form, compact level sets, exhaustion, and the chain to the three Stein properties.

Lean formalization [Intermediate+]

Mathlib does not currently formalise Stein Riemann surfaces or Stein manifolds as first-class objects. A proposed signature, in Lean 4 / Mathlib syntax, sketching the target statement:

[object Promise]

The proof depends on names that do not currently exist in Mathlib (the holomorphic-convexity predicate, strictly plurisubharmonic exhaustions, the Behnke-Stein construction of an exhaustion, Hörmander's $L^{2}$ -theorem in dimension one, the coherent-analytic-sheaf category on a Riemann surface). Each is a candidate Mathlib contribution; until then this unit ships with lean_status: none.

Advanced results [Master]

Stein theory on a Riemann surface is the dimension-one case of a substantive classification in every complex dimension. The general formulation, due to Karl Stein 1951 and developed by Henri Cartan, Jean-Pierre Serre, Hans Grauert, and Lars Hörmander through the 1950s and 1960s, defines a Stein manifold of complex dimension $n$ as a non-compact connected complex manifold $M$ that is holomorphically convex and holomorphically separable, and additionally separates tangent vectors at each point (a condition that is automatic in dimension one). Equivalent characterisations: existence of a strictly plurisubharmonic exhaustion $ϕ : M \to R$ ; vanishing $H^{q} (M, F) = 0$ for $q \geq 1$ and every coherent analytic sheaf $F$ (Theorem B); proper embedding of $M$ as a closed analytic submanifold of some $C^{N}$ . The dimension-one case below is the regime where the classification collapses to a single answer; in dimension $\geq 2$ , non-Stein non-compact manifolds exist and the Stein property becomes a substantive constraint.

The Behnke-Stein exhaustion construction. Forster §25 provides an explicit construction of the strictly subharmonic exhaustion $ϕ : X \to R$ on a non-compact Riemann surface. Choose a Runge exhaustion $X_{1} ⋐ X_{2} ⋐ \dots$ with $⋃_{n} X_{n} = X$ — its existence on every non-compact Riemann surface is the Behnke-Stein topological lemma. On each $X_{n}$ pick a strictly subharmonic boundary-blowing function $ϕ_{n}$ via the Riemann-mapping theorem (06.01.06) and a Green's-function construction; modify by Runge approximation so that $ϕ_{n + 1} - ϕ_{n}$ is uniformly small on $X_{n - 1}$ . The limit $ϕ$ exists, is smooth, is strictly subharmonic everywhere, and is exhausting. The construction is the analytic input to every cohomological vanishing on $X$ — it is the load-bearing object behind Theorem B.

Hörmander's $L^{2}$ -method. Hörmander 1965 $L^{2}$ -estimates and existence theorems for the $\overset{ˉ}{\partial}$ -operator (Acta Math. 113) gives the modern analytic engine for Stein theory. On a Stein manifold $M$ with strictly plurisubharmonic weight $ϕ$ , the $\overset{ˉ}{\partial}$ -equation $\overset{ˉ}{\partial} u = g$ admits a solution $u$ with $L^{2} (ϕ)$ -norm bounded by a multiple of the $L^{2} (ϕ)$ -norm of $g$ , provided $\overset{ˉ}{\partial} g = 0$ . The estimate is the Bochner-Kodaira-Nakano identity applied to the twisted Dolbeault complex. In dimension one, the estimate specialises to the elementary Forster argument via Runge approximation; in higher dimension, it is the only proof that works. The $L^{2}$ -method gives Theorem B as a corollary: every coherent sheaf admits a global $L^{2}$ -solution to the $\overset{ˉ}{\partial}$ -problem on a Stein manifold, hence $H^{q} = 0$ for $q \geq 1$ .

Theorems A and B (Cartan-Serre 1953). Theorem A: on a Stein manifold $M$ , every coherent analytic sheaf $F$ is generated by global sections (the natural map $Γ (M, F) \otimes O_{M} \to F$ is surjective on stalks). Theorem B: on a Stein manifold $M$ , $H^{q} (M, F) = 0$ for every coherent analytic $F$ and every $q \geq 1$ . The two theorems are equivalent by an exact-sequence argument relating Theorem A's surjectivity to Theorem B's vanishing for the kernel sheaf. The proof is by induction on a Stein exhaustion $X_{n} ⋐ X_{n + 1}$ , with a limit-passage argument using Schwartz finiteness and the Runge property of each $X_{n}$ . In dimension one, Forster §29 provides a self-contained proof; in higher dimension, the Cartan-Serre 1953 (CRAS 237) seminar paper is the canonical reference.

Riemann-Roch comparison: compact versus Stein. Compact Riemann surfaces (the 06.04.01 regime) have the Riemann-Roch dimension count $dim H^{0} (L) - dim H^{1} (L) = de g L + 1 - g$ with $H^{1}$ generically non-zero, and Serre duality 06.04.04 to identify $H^{1}$ . On a non-compact Riemann surface, Theorem B kills the right-hand cohomology: $H^{1} (X, L) = 0$ for every line bundle $L$ , the Picard group $Pic (X) = H^{1} (X, O_{X}^{*})$ vanishes, and every line bundle is holomorphically isomorphic to $O_{X}$ . The compact regime stratifies line bundles by genus and degree; the non-compact Stein regime collapses all line bundles to the structure sheaf. The dichotomy compact-vs-Stein on a Riemann surface is the dichotomy generically-nonzero-cohomology-vs-vanishing-cohomology.

Higher-dimensional Stein theory and the Levi problem. A connected non-compact complex manifold $M$ of dimension $n \geq 2$ is Stein iff it admits a strictly plurisubharmonic exhaustion (Grauert 1958). The "iff" carries content: the forward direction is the Levi problem, asking whether a pseudoconvex domain in $C^{n}$ is automatically Stein. Affirmative answer: Oka 1953 (in $C^{2}$ ), Bremermann 1954, Norguet 1954 (in $C^{n}$ ). Grauert 1958 extended to abstract complex manifolds. The Levi problem is the higher-dimensional analogue of "every non-compact RS is Stein"; in dimension one, the Levi problem reduces to the Behnke-Stein theorem with no further work, but in dimension $\geq 2$ it is the central existence theorem of Stein theory.

Stein manifolds and affine schemes. GAGA (Serre 1956 Géométrie algébrique et géométrie analytique) identifies algebraic and analytic categories on a smooth projective variety. The Stein analogue is the comparison between Stein manifolds and affine schemes. Every smooth affine variety over $C$ is Stein in the analytic topology, and every Stein manifold that is biholomorphic to an algebraic affine variety carries an algebraic structure compatible with the analytic one. The general Stein-versus-affine question — when is a Stein manifold algebraisable as an affine variety? — is delicate: Serre's example of a Stein manifold not biholomorphic to any affine variety (Serre 1953) shows the inclusion is strict. The dictionary is sharp on the affine side: affine implies Stein; the converse fails in subtle higher-dimensional cases.

Stein-Weinstein duality. Cieliebak-Eliashberg 2012 From Stein to Weinstein and Back establishes that every Stein manifold $X$ carries a canonical Weinstein structure — a Liouville domain with a Morse function whose critical points are isotropic for the Liouville form. The dictionary is: the strictly plurisubharmonic exhaustion $ϕ$ on $X$ gives the Weinstein Morse function; the symplectic form $ω = - d d^{c} ϕ$ makes $X$ a Liouville domain; the gradient flow of $ϕ$ is the Liouville flow. Conversely every Weinstein domain admits a Stein structure (up to symplectomorphism). The bridge is the central technical input of Floer-theoretic invariants on Stein manifolds: symplectic homology, wrapped Fukaya category, microlocal sheaves on Stein manifolds. The Riemann-surface case of this bridge is degenerate (open Riemann surfaces are 1-dimensional symplectic manifolds, so Weinstein structure reduces to a Morse function with critical points of index $\leq 1$ ), but the higher-dimensional case is a substantive bridge between complex analysis and symplectic topology.

Synthesis. The Behnke-Stein theorem is the structural identification on a Riemann surface: non-compactness implies Stein, and Stein implies the disappearance of every cohomological obstruction in positive degree. The compact-vs-non-compact dichotomy on a Riemann surface is replicated cohomologically as $H^{1} (X, F)$ generically-non-zero vs. uniformly-zero — the entire analytic-spine of complex curves lives in this dichotomy. The Cartan-Serre extension to higher dimension adds a substantive classification: Stein manifolds become a proper subclass of non-compact complex manifolds, separated from the rest by the strictly plurisubharmonic exhaustion or, equivalently, the cohomological vanishing. The Stein property is the analytic incarnation of "the global holomorphic functions on $X$ are abundant enough to do every reasonable thing one would want to do globally" — and on a Riemann surface, this abundance is the automatic consequence of being non-compact. The bridge to symplectic topology (Cieliebak-Eliashberg) and to algebraic geometry (Serre's GAGA + the Stein-versus-affine dictionary) makes Stein theory the meeting point of three distinct geometric universes; on a Riemann surface, the meeting is trivialised, but the structural pattern set down in dimension one is the prototype for every higher-dimensional incarnation.

Full proof set [Master]

Lemma (Radó's theorem). Every connected Riemann surface is second-countable.

Proof. Radó 1925 gives the classical proof: every Riemann surface is paracompact (since complex charts are open subsets of $C$ , locally homeomorphic to $R^{2}$ , and the surface is Hausdorff and connected). Paracompactness plus connectedness on a topological surface implies second-countability by an exhaustion argument with countable refinement. The hypothesis is built into the Stein definition because the Behnke-Stein exhaustion construction needs a countable Runge sequence. $□$

Lemma (Runge exhaustion). Every non-compact connected Riemann surface $X$ admits an exhaustion $X_{1} ⋐ X_{2} ⋐ \dots$ with each $X_{n}$ relatively compact, $⋃_{n} X_{n} = X$ , and each $X_{n}$ Runge in $X$ (every $f \in O (X_{n})$ is uniformly approximable on every compact $K \subset X_{n}$ by functions in $O (X)$ ).

Proof. Pick a countable open cover by relatively compact open discs $D_{1}, D_{2}, \dots$ via paracompactness. Let $X_{1} = D_{1}$ . Inductively define $X_{n + 1}$ as the Runge envelope of $X_{n} \cup D_{n + 1}$ inside $X$ — the smallest Runge open in $X$ containing $X_{n} \cup D_{n + 1}$ . The Runge envelope exists because Runge opens are stable under arbitrary unions, and the Runge envelope of a relatively compact open is relatively compact (this is the Behnke-Stein topological lemma). The sequence is exhausting and Runge by construction. $□$

Lemma (Behnke-Stein exhaustion). Every non-compact connected Riemann surface $X$ admits a strictly subharmonic exhaustion $ϕ : X \to R$ .

Proof. Use the Runge exhaustion $X_{1} ⋐ X_{2} ⋐ \dots$ from the previous lemma. On each $X_{n}$ pick a smooth strictly subharmonic function $ϕ_{n} : X_{n} \to R$ with $ϕ_{n} (z) \to + \infty$ as $z \to \partial X_{n}$ . Construction: pick a smooth real-valued boundary-distance function $d_{n}$ on $X_{n}$ , set $ϕ_{n} = - lo g d_{n}$ near $\partial X_{n}$ and a strictly subharmonic interior model in the centre, smoothed by partition of unity. Modify $ϕ_{n + 1}$ on $X_{n}$ by uniformly approximating the difference $ϕ_{n + 1} - ϕ_{n}$ by a function in $O (X)$ on compact subsets (using the Runge property of $X_{n}$ ). The resulting sequence converges uniformly on compact subsets of $X$ to a smooth limit $ϕ : X \to R$ . By construction $ϕ$ is strictly subharmonic (the strict subharmonicity is open under uniform convergence of $C^{2}$ -functions) and exhausting (the level sets contain the compact sets $ϕ_{n} \leq c$ ). $□$

Lemma ( $\overset{ˉ}{\partial}$ -solvability on a Stein RS). On a non-compact Riemann surface $X$ with strictly subharmonic exhaustion $ϕ$ , the equation $\overset{ˉ}{\partial} u = g$ admits a global $C^{\infty}$ solution $u$ for every $g \in C_{c}^{\infty} (X)$ .

Proof. Hörmander's $L^{2}$ -method in dimension one. Equip $X$ with a Hermitian metric whose Kähler form is $i \partial \overset{ˉ}{\partial} ϕ$ (positive by strict subharmonicity). On weighted $L^{2}$ -spaces with weight $e^{- ϕ}$ , the $\overset{ˉ}{\partial}$ -Laplacian is bounded below by a positive constant (the Bochner-Kodaira-Nakano identity in dimension one reduces to a Cauchy-Schwarz calculation on $∣ g ∣^{2} e^{- ϕ}$ ). Compactness of the support of $g$ gives a finite weighted $L^{2}$ -norm; the $\overset{ˉ}{\partial}$ -Laplacian is then invertible, yielding $u$ with $∥ u ∥_{L^{2} (ϕ)} \leq C ∥ g ∥_{L^{2} (ϕ)}$ . Elliptic regularity ( $\overset{ˉ}{\partial}$ is elliptic in dimension one) bootstraps the $L^{2}$ -solution to a $C^{\infty}$ -solution. The Forster §28 argument is an elementary alternative: explicitly solve the $\overset{ˉ}{\partial}$ -problem on each $X_{n}$ via the Runge property and pass to the limit. $□$

Theorem (Behnke-Stein 1949, full statement). Every connected non-compact Riemann surface $X$ is Stein.

Proof. Using the four lemmas above as packaged inputs: existence of $ϕ$ (Lemma 3), $\overset{ˉ}{\partial}$ -solvability (Lemma 4). The argument proceeds as in the Intermediate-tier proof. Holomorphic convexity: for $K$ compact in $X$ , $\hat{K}$ is contained in a level set $ϕ \leq sup_{K} ϕ$ , hence compact. Holomorphic separability: for $p \neq = q$ , solve a Mittag-Leffler problem at ${p, q}$ via Lemma 4 to construct a holomorphic function with prescribed values $f (p) \neq = f (q)$ . Both Stein conditions hold. $□$

Corollary (Theorem B for Riemann surfaces). On a non-compact Riemann surface $X$ and a coherent analytic sheaf $F$ , $H^{q} (X, F) = 0$ for every $q \geq 1$ .

Proof. The Dolbeault comparison 06.04.02 identifies $H^{q} (X, F)$ with Dolbeault cohomology in $F$ . For $F = O (L)$ a line bundle's section sheaf, Lemma 4 directly produces $H^{1} = 0$ . For higher $q$ , dimensional vanishing on a real two-dimensional surface gives $H^{q} = 0$ for $q \geq 2$ . The general coherent case follows by a finite-presentation reduction (every coherent sheaf admits a local resolution by $O^{k}$ , and the long exact sequence in cohomology together with the line-bundle case gives the general vanishing). $□$

Corollary (Picard-group triviality). On a non-compact connected Riemann surface $X$ , every holomorphic line bundle is isomorphic to the structure-sheaf bundle: $\mathrm{Pic}(X) = H^1(X, \mathcal{O}_X^) = 0$.*

Proof. The exponential exact sequence $0 \to Z \to O_{X} \to O_{X}^{*} \to 0$ gives the long exact sequence

H^{1} (X, Z) \to H^{1} (X, O_{X}) \to H^{1} (X, O_{X}^{*}) \to H^{2} (X, Z) .

The middle term vanishes by the previous corollary, and $H^{2} (X, Z) = 0$ for a non-compact connected surface (no top-dimensional integer cohomology). The map $H^{1} (X, O_{X}^{*}) \to H^{2} (X, Z)$ vanishes, the map from $H^{1} (X, O_{X}) = 0$ is surjective onto $H^{1} (X, O_{X}^{*})$ , hence $H^{1} (X, O_{X}^{*}) = 0$ . $□$

Corollary (Cousin I and Cousin II). On a non-compact Riemann surface $X$ , every Cousin I (additive principal-parts) and Cousin II (multiplicative divisors) datum is realised by a global meromorphic function.

Proof. Cousin I obstruction lies in $H^{1} (X, O_{X}) = 0$ by Theorem B. Cousin II obstruction lies in $H^{1} (X, O_{X}^{*}) = 0$ by the previous corollary. Both vanish. $□$

Connections [Master]

Čech cohomology of holomorphic line bundles 06.04.02. The Stein vanishing $H^{q} (X, F) = 0$ for $q \geq 1$ on a non-compact Riemann surface is the negation of the compact-curve $H^{1}$ generically-non-zero behaviour. The same Čech construction, on the same kind of cover, produces zero in the non-compact regime and a finite-dimensional non-zero space in the compact regime. The Stein theory selects the cover-and-resolution data on which the non-compact vanishing holds.
Holomorphic line bundle on a Riemann surface 06.05.02. On a non-compact connected Riemann surface every holomorphic line bundle is isomorphic to the structure-sheaf bundle (Pic-vanishing corollary above). The line-bundle classification on a Riemann surface is therefore concentrated entirely in the compact case; the non-compact case is the trivialisation of the entire structure.
Riemann-Roch theorem for compact Riemann surfaces 06.04.01. The compact-vs-Stein dichotomy on a Riemann surface is the dichotomy substantive-Riemann-Roch-correction-vs-no-correction. Compact: $dim H^{0} (L) - dim H^{1} (L) = de g L + 1 - g$ with $H^{1}$ generically non-zero. Stein: $H^{1} = 0$ , so $dim H^{0} (L) = de g L + 1 - g$ in any sense the formula extends to. On a non-compact Riemann surface, $dim H^{0} (O_{X}) = \infty$ and the Riemann-Roch identity does not literally apply; the cohomological structure reduces to the Stein vanishing.
Serre duality on a curve 06.04.04. The compact-curve duality $H^{1} (X, L) ≅ H^{0} (X, K_{X} \otimes L^{- 1})^{*}$ is the curve case of a perfect pairing that has no analogue on a non-compact $X$ — both sides of the would-be pairing are infinite-dimensional or zero, and the trace map on $H^{1} (X, K_{X})$ is undefined because the residue theorem fails on a non-compact curve. The Stein vanishing replaces the duality pairing with the simpler statement that the right-hand side of Serre duality also vanishes on a non-compact $X$ (since $H^{0} (X, K_{X} \otimes L^{- 1})$ is infinite-dimensional but $H^{1} (X, L) = 0$ — the would-be isomorphism degenerates).
Meromorphic function 06.01.05. The Mittag-Leffler problem on a non-compact $X$ is always solvable (Cousin I corollary above): every prescribed principal-parts datum at a finite set of points is realised by a global meromorphic function. The classical Mittag-Leffler theorem on $C$ (Mittag-Leffler 1884) is the special case $X = C$ .
Riemann mapping theorem 06.01.06. The Riemann mapping theorem (every simply connected proper open subset of $C$ is biholomorphic to the unit disc) is the Stein-rigidity statement on simply connected non-compact Riemann surfaces: any such surface is biholomorphic to $D$ , $C$ , or $P^{1}$ (uniformisation theorem 06.03.03). The two non-compact cases are Stein; the compact case $P^{1}$ is not.
Uniformization theorem 06.03.03. The non-compact half of uniformisation lists exactly the simply connected Stein Riemann surfaces: $C$ and $H$ (or equivalently $D$ ). Every simply connected Stein Riemann surface is one of these two; the universal cover of any non-compact Riemann surface is one of the two.
Holomorphic functions of several complex variables 06.07.01. The higher-dimensional Stein theory is the substantive classification problem. On $C^{n}$ for $n \geq 2$ , there exist non-Stein non-compact complex manifolds, and the Stein property becomes a condition. On a Riemann surface this is moot.
Hartogs phenomenon 06.07.02. Hartogs' theorem (a function holomorphic on a punctured polydisc in $C^{n}$ , $n \geq 2$ , extends across the puncture) is the obstruction to non-Stein-ness in higher dimension: the punctured polydisc is not Stein, and the extension theorem reflects this. On a Riemann surface there is no analogue — every non-compact Riemann surface is Stein, and there is no Hartogs-type extension obstruction.
Theorems A and B for Stein Riemann surfaces (planned successor 06.09.02). The Cartan-Serre theorems are the cohomological consequences of the Stein property: every coherent analytic sheaf is generated by global sections (A), and has vanishing higher cohomology (B). The Behnke-Stein theorem produces the Stein property; Theorems A and B then produce the entire analytic-spine of non-compact-Riemann-surface theory.

Historical & philosophical context [Master]

Heinrich Behnke and Karl Stein proved the theorem that every non-compact Riemann surface is Stein in 1949 in Entwicklung analytischer Funktionen auf Riemannschen Flächen ^{[Behnke-Stein 1949]} (Math. Ann. 120, 430-461). The paper appeared in the same period as Stein's foundational 1951 Analytische Funktionen mehrerer komplexer Veränderlichen zu vorgegebenen Periodizitätsmoduln und das zweite Cousinsche Problem ^{[Stein 1951]} (Math. Ann. 123, 201-222), where Karl Stein introduced the general Stein-manifold definition in arbitrary complex dimension. The Behnke-Stein paper preceded the Stein-manifold definition by two years, but the theorem is now read in the post-1951 framework as the dimension-one base case of Stein theory.

Henri Cartan and Jean-Pierre Serre proved Theorems A and B in the 1951–53 Cartan séminaire at the École Normale Supérieure ^{[Cartan 1951–53]}, and Cartan-Serre 1953 (CRAS 237) gave the published version. The two theorems unified the Behnke-Stein non-compact-RS theory with the higher-dimensional Stein theory of Stein 1951, and provided the cohomological-vanishing language that has organised every subsequent treatment of Stein manifolds. The Cartan-Serre programme, together with Serre's Faisceaux algébriques cohérents (1955) on the algebraic side, established the coherent-sheaf-cohomology framework as the modern language of complex analytic and algebraic geometry.

Lars Hörmander's 1965 $L^{2}$ -estimates and existence theorems for the $\overset{ˉ}{\partial}$ -operator ^{[Hörmander 1965]} (Acta Math. 113) gave the analytic engine — the $L^{2}$ -method for solving the $\overset{ˉ}{\partial}$ -equation on a strictly plurisubharmonic weighted Hilbert space — that produces Theorem B as a direct corollary on every Stein manifold. In dimension one, Forster §25–§29 ^[Forster] presents the elementary alternative: a Runge-approximation-and-limit-passage argument that avoids weighted $L^{2}$ -spaces and produces the Stein vanishing directly.

Hans Grauert 1958 On Levi's problem and the imbedding of real-analytic manifolds ^{[Grauert 1958]} (Ann. Math. 68) extended the Behnke-Stein theorem to the higher-dimensional Levi problem, proving that every connected non-compact complex manifold with a strictly plurisubharmonic exhaustion is Stein. The dimension-one base case (every non-compact RS is Stein) becomes a substantive classification in dimension $\geq 2$ , with Grauert 1958 providing the existence theorem for plurisubharmonic exhaustions on pseudoconvex domains.

Cieliebak-Eliashberg 2012 From Stein to Weinstein and Back ^{[Cieliebak-Eliashberg 2012]} (AMS Colloquium Publications 59) develops the bridge to symplectic topology: every Stein manifold carries a canonical Weinstein structure, and every Weinstein domain admits a Stein structure compatible with the symplectic form. The bridge is the central technical input of modern symplectic topology on open complex manifolds, and the Riemann-surface case is the degenerate dimension-one base case of the construction.

Bibliography [Master]

[object Promise]

Prerequisites

06.05.02
06.04.02

Used in

06.09.02

Tier anchors

beginner: Forster *Lectures on Riemann Surfaces* §25 informal; Donaldson *Riemann Surfaces* §11 informal
intermediate: Forster *Lectures on Riemann Surfaces* §25–§26; Donaldson *Riemann Surfaces* §11
master: Behnke-Stein 1949 *Entwicklung analytischer Funktionen auf Riemannschen Flächen* (originator: every non-compact RS is Stein); Stein 1951 *Analytische Funktionen mehrerer komplexer Veränderlichen zu vorgegebenen Periodizitätsmoduln und das zweite Cousinsche Problem* (originator: Stein-manifold definition); Cartan 1951–53 séminaire (Theorems A and B); Forster *Lectures on Riemann Surfaces* §25–§26; Hörmander *Several Complex Variables* Ch. IV–VI; Cieliebak-Eliashberg *From Stein to Weinstein and Back* (2012)

References

TODO_REF
Behnke-Stein 1949 — Entwicklung analytischer Funktionen auf Riemannschen Flächen · Math. Ann. 120, 430-461 (originator: every non-compact Riemann surface is Stein)
TODO_REF
Stein 1951 — Analytische Funktionen mehrerer komplexer Veränderlichen zu vorgegebenen Periodizitätsmoduln und das zweite Cousinsche Problem · Math. Ann. 123, 201-222 (originator: Stein-manifold definition)
TODO_REF
Cartan 1951–53 séminaire — Théorèmes A et B · Séminaire Cartan, ENS, exposés 1951–53 (originator of Theorems A and B)
TODO_REF
Forster — Lectures on Riemann Surfaces · Springer GTM 81, §25–§26 (Stein theory and Behnke-Stein on Riemann surfaces)
TODO_REF
Donaldson — Riemann Surfaces · Oxford Graduate Texts 22, §11 (Stein theory on non-compact Riemann surfaces)
TODO_REF
Hörmander — An Introduction to Complex Analysis in Several Variables · North-Holland 1973, Ch. IV–VI ($L^2$ theory and Stein manifolds in several variables)
TODO_REF
Cieliebak-Eliashberg — From Stein to Weinstein and Back · AMS Colloquium Publications 59, 2012 (Stein-Weinstein duality)
TODO_REF
Grauert-Remmert — Theory of Stein Spaces · Springer Grundlehren 236, 1979 (analytic spaces extension)

Reviewer

TBD

Estimated time

beginner: 14m
intermediate: 38m
master: 64m