Abstract
The K-ring of symmetric vector bundles over a scheme $X$, the so-called Grothendieck-Witt ring of $X$, can be endowed with the structure of a (special) $\lambda$-ring. The associated $\gamma$-filtration generalizes the fundamental filtration on the (Grothendieck-)Witt ring of a field and is closely related to the "classical" filtration by the kernels of the first two Stiefel-Whitney classes.
Introduction
In algebraic geometry and K-theory, understanding the structure of rings associated with geometric objects is a fundamental pursuit. One such object is the Witt ring, which classifies quadratic forms (or symmetric bilinear forms) over a field or a scheme. While robust methods exist to filter these rings for fields, extending this to general geometric schemes has historically been a complex challenge.
In this work, Marcus Zibrowius provides an elegant solution by proving that the Grothendieck-Witt ring $GW(X)$ of a scheme is a $\lambda$-ring. This allows for the importation of the $\gamma$-filtration—a powerful tool from K-theory—to provide a natural, intrinsic organization of the structure of Witt rings across all degrees.
The Challenge: Filtering Witt Rings for General Schemes
For a field, the Witt ring possesses a "fundamental filtration" based on powers of the fundamental ideal (the ideal of even-rank forms). This filtration is significant because its graded pieces are isomorphic to Milnor K-theory and Galois cohomology.
However, for a general scheme, no single renowned candidate for a filtration exists. Researchers typically rely on the classical filtration, which is defined only for low degrees using rank and the first two Stiefel-Whitney classes ($w_1, w_2$), or the unramified filtration, which can lose information if the map to the function field is not injective. The objective is to find a filtration that is intrinsic to the scheme, works for all degrees, and remains consistent with the field case.
Methodology: $\lambda$-Rings and Symmetric Functions
The primary methodology involves the theory of $\lambda$-rings. A $\lambda$-ring is a ring equipped with operations that behave like exterior powers ($\Lambda^k$). These operations are fundamentally linked to symmetric functions.
The construction of the $\lambda$-structure on $GW(X)$ involves several steps:
- Establishing the $\lambda$-structure: Exterior power operations on symmetric vector bundles are shown to be well-defined on the Grothendieck-Witt ring.
- Overcoming the Splitting Principle: In standard K-theory, the "splitting principle" simplifies proofs by treating vector bundles as sums of line bundles. Zibrowius demonstrates that this principle fails for $GW(X)$.
- Twisting by Torsors: To circumvent this failure, the author employs representation rings of orthogonal groups and "twisting by torsors" in the étale topology. This proves that the $\lambda$-structure is "special," satisfying specific polynomial identities. This involves universal maps to the ring of Witt vectors $\mathbb{W}A$.
Once the special $\lambda$-structure is established, the $\gamma$-filtration is defined by the standard formula: $\gamma^n(x) = \lambda^n(x + n - 1)$.
Key Insights: Nilpotence and Geometric Information
The identification of $GW(X)$ as a $\lambda$-ring yields several immediate benefits. First, results from $\lambda$-ring theory prove that torsion elements in the ring are nilpotent. Second, the new $\gamma$-filtration is "finer" than the unramified filtration, capturing subtle geometric information that the unramified version might overlook.
Main Theorems and Computational Examples
The paper establishes two primary theorems:
- Theorem 1: For an integral scheme over a field (char $\neq 2$), the $\gamma$-filtration on $GW(k)$ coincides with the fundamental filtration. For degrees 1 and 2, it matches the classical filtration defined by rank and $w_1$. At degree 3, the $\gamma$-filtration is contained within the kernel of $w_2$.
- Theorem 2: Formally proves that exterior power operations give $GW(X)$ the structure of a special $\lambda$-ring for any scheme $X$ over a field (char $\neq 2$).
Computational examples for curves, surfaces, and projective spaces demonstrate where the $\gamma$-filtration and classical filtrations diverge. For instance, for certain surfaces, the third step of the $\gamma$-filtration ($F^3_\gamma$) is strictly smaller than the classical counterpart.
Conclusion and Future Directions
While the $\gamma$-filtration provides a unified framework, discrepancies remain at higher degrees, such as the degree 3 mismatch in specific cases like $C \times \mathbb{P}^1$. Future work may involve extending these results to fields of characteristic 2 or exploring the relationship between the $\gamma$-filtration and higher Stiefel-Whitney classes.
Summary updated