In the study of algebraic geometry, Grassmannians—the spaces parametrizing linear subspaces of a fixed dimension—serve as a fundamental playground for understanding the intersection and containment of geometric objects. A cornerstone of this field is the theory of coisotropic hypersurfaces, developed in the 1990s by Gel’fand, Kapranov, and Zelevinsky (GKZ). However, for decades, this theory was largely restricted to hypersurfaces (codimension one). In their paper, "Isotropic and Coisotropic Subvarieties of Grassmannians," Kathlén Kohn and James Mathews provide a significant generalization of this framework, extending it to subvarieties of any codimension and introducing a dual theory of "isotropic" varieties.
The Problem: Beyond Hypersurfaces
A central tool for studying a projective variety $X$ is its Chow form, which describes the set of all subspaces that intersect $X$. These sets typically form coisotropic hypersurfaces. While the properties of these hypersurfaces were well-documented, a gap remained in the literature regarding cases where these associated sets are not hypersurfaces. Kohn and Mathews address this by defining what it means for a subvariety of a Grassmannian of any codimension to be coisotropic. Furthermore, they explore the previously neglected dual notion—isotropic varieties—based on tangent space properties.
Methodology: Rank Conditions and Incidence Geometry
The authors' approach relies on the identification of tangent and conormal spaces of a Grassmannian with spaces of homomorphisms (which can be viewed as matrices). By analyzing the rank of these homomorphisms, they define "strong" and "weak" versions of coisotropy and isotropy:
- Strongly Coisotropic: A subvariety is strongly coisotropic if every vector in its conormal space has a rank of at most one.
- Coisotropic (Weak): A subvariety is coisotropic if its conormal space is contained in the Zariski closure of spaces spanned by rank-one homomorphisms.
- Isotropy: This is the dual condition applied to the tangent space. A variety is isotropic if its tangent space satisfies similar rank-one conditions.
By utilizing the isomorphism between the Grassmannian $Gr(\ell, \mathbb{P}^n)$ and its dual $Gr(n-\ell-1, (\mathbb{P}^n)^*)$, the authors demonstrate that coisotropy is preserved under projective duality, providing a robust algebraic symmetry to their definitions.
Key Insights and Geometric Characterizations
The Nature of Coisotropy
The paper’s most profound result is Theorem 1, which provides a complete geometric characterization of strongly coisotropic varieties. The authors prove that an irreducible variety is strongly coisotropic if and only if it is an "associated variety" of some projective variety $X$. Specifically, these are the sets of subspaces that meet $X$ non-transversely at a smooth point. This result bridges the gap between abstract rank conditions on conormal spaces and the intuitive geometry of how subspaces intersect varieties.
The Nature of Isotropy
While coisotropy relates to how subspaces intersect a variety, isotropy relates to how subspaces are contained within or contain other fixed spaces. Theorem 11 establishes that for dimensions $\ge 2$, a variety is strongly isotropic if and only if every subspace within it either contains a fixed subspace or is contained within a fixed larger subspace. For the case of curves (dimension 1), the authors provide a classification in Theorem 8, linking isotropic curves to the "osculating spaces" (higher-order tangent spaces) of other curves.
Automatic Properties
Interestingly, the authors find that these properties are sometimes unavoidable. Subvarieties of sufficiently low dimension are "automatically" coisotropic, while those of sufficiently high dimension are "automatically" isotropic. This suggests that these categories are not just niche exceptions but are intrinsic to the structure of Grassmannian subvarieties.
Implications and Future Work
Kohn and Mathews have successfully extended the GKZ framework, but their work opens new avenues for inquiry. While "strong" varieties are now well-characterized, the "weak" versions—those defined by the Zariski closure—remain more mysterious. The authors pose several open questions, including whether weakly coisotropic varieties can be characterized by an underlying variety $X$ and whether the singular loci of coisotropic varieties are themselves coisotropic.
This research lays the theoretical groundwork for advancements in "probabilistic Schubert calculus" and the computational geometry of Chow and Hurwitz forms, providing a more complete map of the subvarieties that populate Grassmannian spaces.
Summary updated