Canonical Trace of Schubert Cycles

• Canonical Trace of Schubert Cycles is defined as the intersection pairing between Schubert varieties and distinguished submanifolds in flag varieties, offering explicit combinatorial formulas based on Weyl group parametrization.
• It yields precise intersection numbers—such as 2^m or 2^(m-1)—depending on the ambient dimension, linking geometric intersections with algebraic singularity profiles.
• The trace formulation extends to determinantal rings by characterizing Gorenstein properties and singular loci through radical trace ideals stable under flat base change.

The canonical trace of Schubert cycles occupies a central position in the intersection theory of flag varieties, singularity theory of determinantal rings, and the study of homological dualities in algebraic geometry. In both geometric and commutative algebraic contexts, the canonical trace encapsulates the intersection pairing of specific cycles—namely, Schubert cycles and distinguished submanifolds or orbits such as the base cycle in flag domains—and encodes structural properties of associated coordinate rings, including their Gorenstein and CTR (canonical–trace–radical) properties. Recent work has yielded explicit combinatorial and algebraic formulas for the canonical trace in both the geometric (intersection-theoretic) and algebraic (ideal-theoretic) settings (Brecan, 2014, Kimura, 26 Jan 2026).

1. Intersection Pairings and the Canonical Trace in Flag Domains

Let $Z = G/B$ be the full flag variety for $G = \mathrm{SL}(n,\mathbb{C})$, with $G_0 = \mathrm{SL}(n,\mathbb{R})$ acting on $Z$. The canonical trace, or intersection pairing, is defined between homology classes corresponding to a Schubert variety $S \subset Z$ and a closed $K$-orbit $C_0$ (where $K \cong \mathrm{SO}(n,\mathbb{C})$ is the complexified maximal compact subgroup). When $S$ and $C_0$ meet transversally, the pairing

$$\langle [S], [C_0] \rangle_{\mathrm{can}} := \#(S \cap C_0)$$

counts the number of isolated, transverse intersection points (Brecan, 2014). The collection of all such classes $[S_w]$ for $w$ indexing Schubert varieties of dimension $\dim S_w = \dim Z - \dim C_0$ with nonempty $S_w \cap C_0$ yields a basis for the relevant homology group.

The explicit homological identity takes the form: $[C_0] = \sum_{w} \#(S_w \cap C_0)[S_w],$ where the sum is over $w$ such that $S_w$ has complementary dimension and intersects $C_0$ nontrivially.

2. Schubert Cycles, Weyl Group Parametrization, and Double-Box Contraction

Schubert cycles in $Z = G/B$ are indexed by the Weyl group $W = S_n$, with each $w \in W$ determining a closure of a $B$-orbit $S_w$. Not all $S_w$ meet $C_0$ or have the correct dimension to participate in the canonical trace pairing. The indexing $w$ for such cycles must satisfy the “double-box-contraction” condition: $w$ must be constructed by a successive "immediate-predecessor" algorithm on indices, i.e., $w = k_1 \ldots k_m l_* l_m \ldots l_1$ (for $n=2m$ or $2m+1$), where each pair $(l_i, k_i = l_i - 1)$ is chosen consecutively (Brecan, 2014).

This condition singles out the collection $\mathcal{S}_{C_0}$ of Schubert varieties that both have the required complementary dimension and actually meet $C_0$. The total number of such cycles is precisely $n!! = (n-1)(n-3)\cdots 1$ (the double factorial).

3. Intersection Numbers and Explicit Formulas

For Schubert varieties $S_w$ satisfying the double-box-contraction condition and the complementary dimensionality, the pairing $\langle [S_w], [C_0] \rangle_{\mathrm{can}}$ takes on explicit values dependent on $n$:

• If $n = 2m+1$, the intersection number is $2^m$,
• If $n = 2m$, the intersection number is $2^{m-1}$.

The base cycle class expands as

$$[C_0] = \begin{cases} 2^m \sum_{w \in \mathcal{S}_{C_0}} [S_w], & \text{if } n=2m+1, \ 2^{m-1} \sum_{w \in \mathcal{S}_{C_0}} [S_w], & \text{if } n=2m, \end{cases}$$

within $H^*(Z)$, with $[S_w]$ varying over the double-box-contraction permutations. A perfect duality is established: for a dual base cycle $C_{w'}$, $\#(S_w \cap C_{w'}) = \delta_{w,w'}$, and thus the intersection matrix is the identity (Brecan, 2014).

4. Algebraic Formulation: Canonical Trace in Determinantal Rings

In the commutative algebraic context, let $B$ be a Noetherian ring, $X = (X_{ij})$ an $m \times n$ matrix of indeterminates, with $G_B(X;\lambda)$ the homogeneous coordinate ring of the Schubert cycle associated to a strictly increasing $m$–tuple $\lambda = [a_1 < \cdots < a_m]$ (Kimura, 26 Jan 2026). The module-theoretic canonical trace refers to the trace ideal of the canonical module $\omega_G$ in $G = G_B(X;\lambda)$.

For $B$ a Gorenstein normal domain, $G$ is Cohen–Macaulay. The divisor class $[\omega_G]$ is determined by the combinatorics of "blocks and gaps" within $\lambda$, leading to explicit representative ideals in the divisor class group. The trace $\Tr(\omega_G)$ is computed as an intersection: $\Tr(\omega_G) = \prod_{h=1}^{\Delta'} \left( \bigcap_{u \in U_h} J_G(X; \sigma_u) \right),$ where $U_h$ is an explicit set of indices determined from the blocks-and-gaps decomposition, and each $J_G(X; \sigma_u)$ is an ideal of minors (details above). The minimal primes of $\Tr(\omega_G)$ are the codimension-one Schubert-primes $J_G(X;\sigma_u)$ (Kimura, 26 Jan 2026).

5. Non-Gorenstein Locus and the CTR Property

The support of the trace ideal,

$V(\Tr(\omega_G)) = \bigcup_{u \in \cup_h U_h} V(J_G(X;\sigma_u)),$

coincides with the non-Gorenstein locus of $G$—that is, a prime $\mathfrak{p}$ has $G_\mathfrak{p}$ Gorenstein if $\mathfrak{p}$ does not contain any of the Schubert-primes $J_G(X;\sigma_u)$. This provides an explicit algebraic characterization of singular loci within Schubert cycles, compatible with established descriptions in the literature on determinantal and Schubert singularities (Kimura, 26 Jan 2026).

A Cohen–Macaulay ring $R$ is \textit{canonical-trace-radical} (CTR) if the trace ideal of its canonical module is radical at every localization $R_\mathfrak{p}$. For $G_B(X;\lambda)$, CTR holds if and only if $B$ is CTR and $\Delta' \le 1$. In these cases, the trace ideal simplifies to a finite intersection of relevant primes and is preserved under arbitrary flat base change.

6. Explicit Examples and Computation

For concrete computation, consider the Schubert cycle in $\mathrm{Gr}(2,5)$ determined by $\lambda = [1,4]$. The blocks-and-gaps decomposition yields explicit sets $b_0, \chi_0, b_1, \chi_1$, and the trace formula produces $\Tr(\omega_G) = J_G(X; [1,5])$. Here, $G$ fails to be Gorenstein precisely along this prime, illustrating the geometric–algebraic correspondence in the singularity structure of Schubert cycles (Kimura, 26 Jan 2026).

Explicit enumeration of intersection points in the flag domain context proceeds via canonical forms of flags with prescribed coordinates, with sign choices parametrizing the intersection points, matching the predicted intersection numbers from the combinatorial theory (Brecan, 2014).

7. Broader Implications and Preservation Under Base Change

All trace formulas and characterizations above are stable under flat base change $B \to C$, as the relevant algebraic and combinatorial structures (ASL structures on coordinate rings, blocks-and-gaps, and minors) are compatible with such extensions. This provides flexibility for passing between characteristic zero and positive characteristic, as well as constructing families of Schubert cycles over varying base rings relevant in moduli or deformation-theoretic contexts (Kimura, 26 Jan 2026).

The canonical trace of Schubert cycles thus bridges intersection theory, combinatorics of Weyl groups, commutative algebra of determinantal and Plücker rings, and singularity theory, with explicit combinatorial control over both geometric intersection numbers and algebraic singular loci.