BGG Sequences -- A Riemannian perspective

Technical Summary: BGG Sequences — A Riemannian Perspective

1. Problem Statement
Bernstein-Gelfand-Gelfand (BGG) sequences are fundamental structures in representation theory and differential geometry. Originally arising from resolutions of Verma modules for semisimple Lie algebras, they were generalized to "parabolic geometries" (curved geometric structures modeled on homogeneous spaces $G/P$). These sequences consist of invariant differential operators acting on sections of natural vector bundles.

However, a significant gap exists between the highly abstract, technical machinery required for general parabolic geometries (involving Cartan connections, tractor bundles, and complex representation theory) and the concrete, applied mathematics context of domains in $\mathbb{R}^n$ (where flat metrics allow for simpler constructions).

• The Challenge: Existing constructions for curved manifolds rely heavily on the theory of Cartan geometries and tractor connections, which are technically demanding and obscure the explicit form of the operators. Conversely, constructions for flat domains (e.g., in elasticity theory) often lack the geometric generalization to curved Riemannian manifolds or rely on ad-hoc methods without representation-theoretic foundations.
• The Goal: The paper aims to bridge this gap by constructing BGG sequences on general Riemannian manifolds (specifically for conformal and projective structures) without invoking the full machinery of parabolic geometries. The goal is to provide an explicit, intermediate-level construction that is accessible to differential geometers and applicable to numerical analysis in applied mathematics.

2. Methodology
The author, Andreas Čap, employs a "Riemannian perspective" that simplifies the general parabolic geometry framework by fixing a specific geometric structure (a metric or a volume-preserving connection) rather than working with the full conformal or projective class.

• Algebraic Setup: The construction relies on the representation theory of the orthogonal group $O(n)$ (for conformal cases) and the special linear group $SL(n, \mathbb{R})$ (for projective cases). These are viewed as subgroups of the larger groups $O(n+1, 1)$ and $SL(n+1, \mathbb{R})$, respectively.
• Bundle Decomposition: An irreducible representation $V$ of the larger group is decomposed into a graded sum $V = \bigoplus V_j$ based on the action of a grading element. This induces a decomposition of the associated vector bundle $V_M$ on the manifold.
• Twisted Connections: Instead of using the abstract tractor connection, the author defines a twisted connection $\nabla^V$ on $V_M$ explicitly using the Levi-Civita connection (or the given linear connection) and the Schouten tensor $P$:
  $$\nabla^V_\xi s = \nabla_\xi s + \xi \bullet s - P(\xi) \bullet s$$
  Here, $\bullet$ denotes a tensorial operation derived from the Lie algebra action.
• Twisted de Rham Sequence: The twisted connection is coupled with the exterior derivative to form a sequence of differential operators $d_{\nabla^V}$. Unlike the flat case, this sequence is generally not a complex (i.e., $d_{\nabla^V} \circ d_{\nabla^V} \neq 0$) unless the underlying geometry is flat (conformally or projectively flat).
• Cohomology Bundles and Splitting: The construction utilizes the operator $\partial$ (induced by the Lie algebra differential) to define cohomology bundles $H^k_V M = \ker(\partial) / \text{im}(\partial)$. A "splitting operator" $S$ is constructed to lift sections from these cohomology bundles back to the full bundle $V_M$.
• Compression: The BGG operators $D$ are obtained by "compressing" the twisted de Rham sequence. Specifically, $D$ is defined as the projection of the twisted derivative of the split section onto the next cohomology bundle: $D(\alpha) = \text{proj}(d_{\nabla^V}(S(\alpha)))$.

3. Key Contributions

• Explicit Construction without Tractors: The paper provides a self-contained construction of BGG sequences on Riemannian manifolds that avoids the abstract theory of Cartan connections and tractor bundles. It replaces them with explicit formulas involving the Schouten tensor and standard covariant derivatives.
• Unified Framework for Conformal and Projective Cases: The methodology is applied to two major cases:
  1. Conformal BGG Sequences: Defined on Riemannian manifolds, utilizing the Schouten tensor derived from the metric.
  2. Projective BGG Sequences: Defined on manifolds with a volume-preserving linear connection, utilizing the projective Schouten tensor.
• Handling Curvature: The paper explicitly addresses the non-flat case. It demonstrates that while the twisted sequence is not a complex in curved geometries, the BGG sequence still provides a valid resolution of parallel sections (in the flat case) or a system of overdetermined PDEs (in the curved case).
• Universal Formulas: The author derives universal formulas for BGG operators of order 1 and 2 (Theorem 2.9), showing how they depend on the location of the cohomology bundles within the graded decomposition.

4. Key Results

• Theorem 2.2 & 3.2 (Curvature): The curvature of the twisted connection $\nabla^V$ is explicitly given by $R^V(\xi, \eta) = W(\xi, \eta) \bullet s + Y(\xi, \eta) \bullet s$, where $W$ is the Weyl curvature and $Y$ is the Cotton-York tensor. The connection is flat if and only if the underlying geometry is conformally/projectively flat.
• Theorem 2.7 (Flat Case): If the geometry is flat, the BGG sequence is a fine resolution of the sheaf of local parallel sections of the tractor bundle, computing the same cohomology as the twisted de Rham sequence.
• Theorem 2.8 (Curved Case): In the general curved case, the kernel of the first BGG operator contains the space of parallel sections of the twisted connection (specifically, "normal solutions"). This establishes a link between the BGG sequence and the geometric prolongation of overdetermined systems.
• Explicit Examples:
  • Conformal Case: The first BGG operator for the standard representation yields the conformal Killing operator. The splitting operator $S$ is explicitly computed, relating solutions to Einstein metrics.
  • Projective Case: The construction recovers the Riemannian deformation sequence (Calabi complex) and the Killing operator on forms. The first BGG operator for the dual standard representation is shown to be related to Ricci-flat connections.
• Order of Operators: The paper clarifies that the order of the BGG operators is determined by the grading shift. First-order operators arise when the cohomology bundle stays in the same grading level, while second-order operators arise when the level shifts.

5. Significance

• Bridging Theory and Application: This work is crucial for applied mathematics, particularly in elasticity theory, numerical relativity, and finite element exterior calculus. By providing explicit formulas for BGG sequences on curved manifolds, it enables the development of numerical methods (e.g., finite element methods) for studying operators relevant to these fields without requiring the full machinery of parabolic geometry.
• Simplification of Geometric Analysis: It demystifies the BGG machinery, making it accessible to researchers who may not specialize in representation theory or Cartan geometry. It shows that the core ideas can be implemented using standard Riemannian tools (Levi-Civita connection, Schouten tensor).
• Generalization of Flat Results: It successfully extends the "finite element exterior calculus" results from flat domains in $\mathbb{R}^n$ to general Riemannian manifolds, preserving the structural properties of the complexes while accounting for curvature terms.
• Foundation for Future Work: The paper sets the stage for further interaction between the differential geometry and applied mathematics communities, offering a common language and explicit tools for analyzing overdetermined systems of PDEs on curved spaces.