🔢 mathematics

Extension and lifting of G-bundles on stacks

This paper investigates extension properties for morphisms of stacks of bundles over group algebraic spaces, applying these results to provide a concise classification of bundles on the projective line for smooth geometrically reductive groups and to establish the existence of splittings for filtered fiber functors of gerbes bound by quasi-affine smooth group schemes over rings.

Original authors: Torsten Wedhorn

Published 2026-08-18

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Torsten Wedhorn

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where shapes are not just drawn on paper but are built from the very rules of symmetry that govern how things can be moved, rotated, or transformed without breaking their essential nature. In the branch of mathematics known as algebraic geometry, researchers study these shapes, which they call schemes, and the bundles of symmetry that can be wrapped around them. Think of a bundle not as a package of goods, but as a way of attaching a specific type of symmetry to every point on a shape, creating a complex structure that holds together only if the rules of symmetry are perfectly consistent. The central question for mathematicians working in this field is often about extension: if you have a shape with a hole in it, and you have successfully wrapped a bundle of symmetry around the part that exists, can you always stretch that wrapping to cover the hole? Sometimes the answer is yes, and sometimes the rules of symmetry are so rigid that the wrapping simply cannot be extended without tearing or changing its fundamental character.

Torsten Wedhorn, in a recent study, tackles this question of extension and lifting for these symmetry bundles, but he does so in a setting that is far more abstract and flexible than the usual shapes found in textbooks. He works with objects called stacks, which are a sophisticated generalization of geometric spaces that allow mathematicians to handle situations where points might have hidden symmetries or where the space itself is built from pieces that overlap in complicated ways. The paper focuses on a specific type of symmetry group, one that is smooth and affine, meaning it behaves nicely and can be described by equations, and it asks when a bundle defined on a smaller part of a stack can be uniquely extended to the whole stack, or when a bundle defined on a slightly larger space can be pulled back to a smaller one without losing information. The research provides a clear set of conditions under which these extensions are guaranteed to work, offering a powerful new tool for understanding how these complex geometric structures fit together.

The core of the work involves proving that under certain reasonable conditions, the act of restricting a bundle to a smaller piece of a space is a perfect match for the act of extending it back. Specifically, the author shows that if you have a map between two such spaces that is well-behaved and preserves the basic structure of the space, then any bundle of symmetry on the smaller space can be extended to the larger one, and this extension is unique. This is not just a theoretical curiosity; it solves a long-standing problem regarding the classification of these bundles on a specific, fundamental shape known as the projective line. For decades, mathematicians had to rely on heavy machinery and complicated stratifications to classify these bundles, but Wedhorn's approach bypasses those difficult arguments entirely. By using a clever diagram that relates the projective line to a simpler space involving a multiplicative group, the paper demonstrates that the classification of bundles on the projective line is exactly the same as the classification of bundles on a much simpler, one-dimensional object. This equivalence means that the complex behavior of bundles on the projective line can be understood by studying the much more manageable behavior of bundles on a line with a specific type of symmetry attached to it.

Beyond the classification of bundles on the projective line, the paper also addresses a different but related problem concerning the "lifting" of bundles. This involves a scenario where a space and a smaller, closed part of that space form what is called a henselian pair, a technical term that essentially means the smaller part is so tightly integrated into the larger space that any symmetry bundle defined on the small part can be extended to the whole. The author proves that for a wide range of symmetry groups, this lifting is always possible and unique. This result has a direct and surprising application to the theory of fiber functors, which are mathematical tools used to translate between different types of algebraic structures. The paper shows that any "filtered" fiber functor, which is a structure that organizes information in layers, can be split into a "graded" version, where the layers are separated cleanly. This splitting was previously known only in very specific cases, such as when the underlying field is a number field, but Wedhorn's work proves it holds for a much broader class of rings and groups.

The significance of these findings lies in their generality and the clarity of the methods used. The author does not rely on the specific properties of the projective line or the particular nature of the groups involved in the way previous proofs did. Instead, the paper establishes general principles about how bundles behave when moving between different types of spaces. These principles are robust enough to apply to groups that are not necessarily connected, a situation that had previously been difficult to handle. The results confirm that the classification of bundles on the projective line over any field is determined entirely by the cocharacters of the group, which are essentially ways of mapping a simple multiplicative group into the symmetry group. This provides a unified and elegant description of these bundles, replacing a patchwork of special cases with a single, coherent framework.

In the end, the paper offers a new perspective on how symmetry and geometry interact. By proving that certain extension and lifting problems have definitive solutions under broad conditions, it removes a layer of uncertainty that had persisted in the field. The work demonstrates that the complex world of algebraic stacks, which can seem impenetrable to those outside the field, follows logical rules that allow for precise predictions about the behavior of bundles. The ability to classify bundles on the projective line and to split filtered fiber functors is not just a technical victory; it opens the door to further exploration of how these structures behave in even more complex settings. The paper stands as a testament to the power of abstract reasoning, showing that by stepping back and looking at the general properties of these mathematical objects, one can find simple, universal truths that govern their behavior.

Technical Summary: Extension and lifting of G-bundles for stacks

Problem Statement
The paper investigates extension and lifting properties for morphisms of $G$ -bundles, where $G$ is a group scheme (or group algebraic space) and the base objects are stacks (not necessarily algebraic). The central inquiry concerns the behavior of the pullback functor $f^*: \text{Bun}_G(X) \to \text{Bun}_G(Y)$ induced by a morphism of stacks $f: Y \to X$ . Specifically, the author seeks criteria under which:

$f^*$ is fully faithful or an equivalence of categories (Extension problem).
$f$ is a closed immersion and $f^*$ induces a surjection or bijection on isomorphism classes (Lifting problem).

The motivation includes providing a new, streamlined proof for the classification of $G$ -bundles on the projective line $\mathbb{P}^1_k$ for affine smooth group schemes with reductive identity components, and establishing the existence of splittings for filtered fiber functors.

Methodology
The paper employs a combination of descent theory, stack geometry, and topological arguments regarding connected components.

Extension via Full Faithfulness: The author establishes that if $f: U \to X$ is a quasi-compact and quasi-separated (qcqs) morphism representable by schemes satisfying $\mathcal{O}_X = f_*\mathcal{O}_U$ , then the restriction functor on vector bundles and $G$ -bundles is fully faithful (Proposition 2.4). To address essential surjectivity (extension), the paper utilizes an embedding technique: if $G$ embeds into a flat affine group $H$ such that the quotient $G \backslash H$ is affine, the extension problem for $G$ -bundles reduces to that of $H$ -bundles (Proposition 2.5). This allows the application of known results for vector bundles (e.g., Auslander–Buchsbaum theorem for regular schemes of dimension 2).
Lifting via Henselian Pairs: The lifting problem is addressed by defining "henselian pairs of algebraic stacks." The author first proves a purely topological result (Proposition 3.3) stating that for spectral spaces, a continuous map induces a bijection on open and closed subsets if and only if it induces a bijection on connected components. This topological characterization is used to define henselian pairs of stacks (Definition 3.7).
Quotient Stacks and Fixed Points: The core lifting theorem (Theorem 3.11) applies to pairs of quotient stacks $([H \backslash \tilde{X}], [H \backslash \tilde{Z}])$ where $H$ is a group of multiplicative type. The result relies on a theorem by Alper, Hall, and Rydh [3] regarding the lifting of morphisms from gerbes. The condition for the pair to be henselian is reduced to the fixed-point loci $(\tilde{X}^H, \tilde{Z}^H)$ forming a henselian pair.
Graded Algebra Reformulation: In Section 3.3, the lifting theorem is reformulated in terms of graded algebras. For a graded algebra $B$ concentrated in non-negative degrees, the pair $([G_m \backslash \text{Spec } B], BG_{m, B_0})$ is shown to be henselian, facilitating the lifting of bundles from the fixed point locus (the "zero section") to the total space.

Key Contributions and Results

General Extension Criteria (Proposition 2.4, 2.5): The paper provides general conditions for the full faithfulness of the restriction functor for $G$ -bundles on stacks. It establishes that if $G$ is flat and affine, and $\mathcal{O}_X = f_*\mathcal{O}_U$ , restriction is fully faithful. Furthermore, it provides a criterion for essential surjectivity via embedding $G$ into a larger group $H$ with an affine quotient.
Lifting Theorem for Henselian Pairs (Theorem 3.11): The main technical result states that for a group $H$ of multiplicative type (diagonalizable after finite locally free base change) acting on an affine scheme $\tilde{X}$ , if the fixed-point pair $(\tilde{X}^H, \tilde{Z}^H)$ is henselian, then the restriction functor on $G$ -bundles (for smooth quasi-affine $G$ ) from the quotient stack $[H \backslash \tilde{X}]$ to $[H \backslash \tilde{Z}]$ is essentially surjective and full. This yields a bijection on isomorphism classes $H^1([H \backslash \tilde{X}], G) \cong H^1([H \backslash \tilde{Z}], G)$ .
Classification of Bundles on $\mathbb{P}^1_k$ (Theorem 4.2): The paper proves a natural bijection $H^1(\mathbb{P}^1_k, G) \cong H^1(B\mathbb{G}_{m,k}, G)$ $H^{1} (P_{k}^{1}, G) ≅ H^{1} (B G_{m, k}, G)$ for any affine smooth group scheme $G$ $G$ over a field $k$ $k$ whose identity component is reductive.
- Methodology: The proof bypasses the Harder–Narasimhan stratification and vector bundle classification. Instead, it uses the diagram $\mathbb{P}^1_k = [\mathbb{G}_m \backslash (\mathbb{A}^2 \setminus \{0\})] \hookrightarrow [\mathbb{G}_m \backslash \mathbb{A}^2] \leftarrow [\mathbb{G}_m \backslash \{0\}] = B\mathbb{G}_m$ . The map from $\mathbb{A}^2 \setminus \{0\}$ to $\mathbb{A}^2$ induces a bijection on bundles (via Auslander–Buchsbaum), and the map from the origin to $\mathbb{A}^2$ induces a bijection via the lifting theorem for henselian pairs.
- Significance: This offers a new proof for the classification of Zariski locally trivial $G$ -bundles over $\mathbb{P}^1_k$ for reductive groups, generalizing previous results by Harder, Biswas, Nagaraj, and Anschütz to non-connected groups.
Splitting of Filtered Fiber Functors (Corollary 4.7): The paper proves that every filtered fiber functor of $\text{Rep}(G_0)$ $Rep (G_{0})$ over an affine scheme $S$ $S$ admits a splitting (i.e., is isomorphic to a graded fiber functor) if $G$ $G$ is smooth and quasi-affine.
- Context: This generalizes a theorem of Ziegler [33] from fields to general rings. The result is derived by identifying the stack $[\mathbb{G}_m \backslash \mathbb{A}^1_S]$ with the stack of filtered fiber functors and applying the lifting theorem to the closed immersion of the zero section.

Significance and Claims
The author claims that the primary interest of Theorem 1 (the classification on $\mathbb{P}^1_k$ ) lies not merely in the generalization to non-connected groups, but in the method of proof. By avoiding the Harder–Narasimhan stratification, the paper demonstrates that extension and lifting properties of bundles on stacks can be derived directly from the geometry of the underlying quotient stacks and henselian pairs.

The paper asserts that the general results on extension (Proposition 2.4/2.5) and lifting (Theorem 3.11) appear to be new in this level of generality, even if they are not surprising in specific cases. The lifting theorem is presented as a generalization of results by Bouthier and Česnavičius [7] to the context of quotient stacks by groups of multiplicative type.

The work also provides a new characterization of henselian pairs of algebraic stacks via connected components (Proposition 3.6) and explicitly connects the splitting of filtered fiber functors to the geometric properties of the stack $[\mathbb{G}_m \backslash \mathbb{A}^1]$ . The paper notes that the smoothness hypothesis on $G$ is necessary for the uniqueness of the splitting, citing counterexamples for non-smooth groups.

More like this