Canonical bases and collective integrable systems

Benjamin Hoffman, Jeremy Lane

Abstract

Let $K$ be a non-abelian compact connected Lie group. We show that every Hamiltonian $K$-manifold $M$ admits a Hamiltonian torus action of the same complexity on a connected open dense subset. The moment map for this torus action extends continuously to all of $M$. This generalizes the celebrated Gelfand-Zeitlin integrable systems introduced by Guillemin and Sternberg. In the process, we develop a general framework for integrating gradient Hamiltonian vector fields on degenerations. This framework can be applied to degenerations of varieties that are possibly singular or non-compact. It recovers a result of Harada and Kaveh as a special case. Our main result on Hamiltonian $K$-manifolds is obtained by applying this framework to the base affine space of the complexification of $K$.

Introduction: The Challenge of Non-Abelian Integrability

In the study of Hamiltonian mechanics, integrable systems are highly valued for their ability to describe complex motion through independent conserved quantities. For decades, the Gelfand-Zeitlin (GZ) system has served as a primary tool for analyzing manifolds with symmetries from the unitary group $U(n)$ or orthogonal group $SO(n)$.

However, a significant gap persisted: GZ systems did not naturally extend to other symmetry groups, such as the symplectic group $Sp(n)$ or exceptional groups. Previous attempts to generalize these systems often failed to cover all group orbits or lost essential geometric features, such as the convexity of the moment map image and the connectedness of its fibers. Furthermore, existing toric degeneration techniques typically required smooth, projective spaces, leaving singular or non-compact varieties unaddressed.

Conceptual visualization of a Hamiltonian flow between states.

Methodology: Stratified Gradient Hamiltonian Flows

The authors bridge this gap by introducing a novel framework: Stratified Gradient Hamiltonian Vector Fields. This approach allows for the integration of flows even in the presence of singularities by decomposing the space into smooth "strata."

  1. Toric Degeneration: The variety representing the symmetry group is algebraically degenerated into a simpler toric variety.
  2. Stratification: Because these varieties are often singular, the space is decomposed into smooth pieces that fit together.
  3. Piece-wise Integration: A gradient Hamiltonian vector field is defined on each smooth stratum. The technical breakthrough lies in proving that the flow of this piece-wise defined field is continuous and has a well-defined limit as it approaches the simplified toric version of the space.
  4. Valuations: By utilizing "good valuations" on the coordinate ring of the base affine space, the authors construct a universal integrable system that applies to any manifold the group acts upon.
Integration of gradient Hamiltonian vector fields on stratified degenerations.

Key Insights: Universality and Complexity Preservation

  • Universal Applicability: By focusing on the base affine space of the complexified Lie group, the authors create a "master" integrable system. Any manifold with a Hamiltonian action of group $K$ can pull back this system to generate its own torus action.
  • Complexity Preservation: A major finding is that the complexity of the original non-abelian action is identical to the complexity of the new torus action. For multiplicity-free actions, the resulting torus action is completely integrable.
  • Geometric Consistency: This method ensures the moment map image is a rational convex polytope, connecting the work to the Kirwan polytope and Kashiwara-Lusztig crystal bases in representation theory.

Results: Generalizing Gelfand-Zeitlin Systems

The primary result is the proof of existence for a completely integrable system for any compact connected Lie group $K$ (Theorem 1.1). These systems maintain connected fibers and convex images, overcoming the limitations of previous generalizations.

Furthermore, the paper provides a path toward solving the Karshon-Tolman conjecture regarding the Gromov width of coadjoint orbits. By eliminating technical roadblocks for "irrational" orbits, the authors enable calculations across all Lie types. They also successfully construct graded isomorphisms between Bohr-Sommerfeld fibers and Kähler quantization bases, a significant result for geometric quantization.

Future Work

While the moment map extends to the entire manifold, the Hamiltonian torus action is currently guaranteed only on a connected open dense subset. The behavior on the manifold's boundary remains a complex area for future study. Additionally, the application to Gromov width rests on an algebraic conjecture regarding "good birational orderings," the proof of which for all Lie types is a clear direction for subsequent research.

Summary updated

Abstract

Let K be a non-abelian compact connected Liegroup. We show that every Hamiltonian K-manifold M admits a Hamiltonian torus action of the same complexity on a connected open dense subset. The moment map for this torus action extends continuously to all of M . This generalizes the celebrated Gelfand-Zeitlin integrable systems introduced by Guillemin and Sternberg. In the process we develop a general framework for integrating gradient Hamiltonian vector fields on degenerations. This framework can be applied to degenerations of varieties that are possibly singular or non-compact. It recovers a result of Harada and Kaveh as a special case. Our main result on Hamiltonian K-manifolds is obtained by applying this framework to the base affine space of the complexification of K.
BibTeX
@article{hoffman2025canonical,
  author = {Benjamin Hoffman and Jeremy Lane},
  title = {{Canonical bases and collective integrable systems}},
  journal = {Journal of the European Mathematical Society},
  year = {2025},
  doi = {10.4171/jems/1659},
}
Figure