Stein spaces and Stein algebras

We prove that the category of Stein spaces and holomorphic maps is anti-equivalent to the category of Stein algebras and $\mathbb{C}$-algebra morphisms. This removes a finite dimensionality hypothesis from a theorem of Forster.

💡 Research Summary

The paper establishes a full anti‑equivalence between the category of Stein spaces (with holomorphic maps) and the category of Stein algebras (with complex‑algebra morphisms), removing the finite‑dimensionality hypothesis that was present in Forster’s classical results. After recalling the definitions of Stein spaces and Stein algebras—complex spaces whose global holomorphic function algebras carry the Fréchet topology—the author notes that Forster proved continuity of algebra morphisms and the anti‑equivalence only for finite‑dimensional Stein spaces. Subsequent work by Markoe and Ephraim weakened the dimensional assumptions but did not eliminate them entirely.

The core of the new argument is Theorem 2.1, which asserts that for any Stein space (S) there exists a holomorphic map (f:S\to\mathbb C^{2}) whose fibers are all finite‑dimensional. The construction relies on recent advances in Oka theory. The author introduces a family of Oka manifolds (Y_{r}\subset\mathbb C^{2}) (originally studied by Förster–Wold) and proves they are Oka and contractible. Using the Oka extension theorem (Förster 2005, 2017), any holomorphic map defined on a closed complex subspace of (S) with values in (Y_{r}) can be extended to the whole of (S). By organizing the irreducible components of (S) into a well‑ordered countable set (\Theta) and proceeding inductively, compatible holomorphic maps are built on increasing unions of components, each time ensuring that the image of the new component lies in a suitable (Y_{r}). The Oka property guarantees the existence of extensions, and the decreasing family ((Y_{k})_{k\ge0}) with empty intersection forces each fiber of the final map (f) to intersect only finitely many components, hence to be finite‑dimensional. Proposition 2.3 shows this result is optimal: there exists a Stein space for which every holomorphic map to (\mathbb C) has an infinite‑dimensional fiber.

With such a map (f) in hand, Theorem 3.1 proves that any (\mathbb C)-algebra morphism (\chi:\mathcal O(S)\to\mathbb C) is automatically continuous and must be evaluation at a point of (S). The proof evaluates (\chi) on the coordinate functions of (f), defines a closed subspace (T\subset S) where these coordinates equal the corresponding complex numbers, and observes that (T) is finite‑dimensional. Forster’s original continuity theorem for finite‑dimensional Stein spaces then applies, yielding continuity of (\chi); the subsequent Forster result (Satz 1) gives the evaluation representation. Theorem 3.2 extends this to arbitrary morphisms between Stein algebras, showing they are continuous by reducing to the case of maximal ideals and invoking the continuity of evaluation maps.

Finally, Theorem 3.3 combines these ingredients to obtain the desired anti‑equivalence: the contravariant functor (S\mapsto\mathcal O(S)) is fully faithful and essentially surjective on objects, because every Stein algebra arises as the global sections of a Stein space and every morphism of Stein algebras is continuous and thus corresponds to a holomorphic map. Proposition 3.4 records a useful corollary: for any complex space (X) and Stein space (S), holomorphic maps (X\to S) correspond bijectively to morphisms of locally ringed spaces (X\to\operatorname{Spec}(\mathcal O(S))).

In summary, by leveraging modern Oka theory—particularly the existence of Oka domains in Euclidean spaces and extension theorems—the author removes all dimensional restrictions from Forster’s foundational results, thereby providing a complete categorical duality between Stein spaces and their function algebras. This advances the parallel between complex analytic and algebraic geometry, aligning the analytic theory of Stein spaces with the well‑known algebraic correspondence between affine varieties and finitely generated (\mathbb C)-algebras.