Concurring reduction schemes for Dirac structures

Abstract: The notion of \emph{concurrence} was recently proposed as the natural compatibility relation between Dirac structures, generalizing the commutativity of two Poisson structures. We address the question of when a reduction scheme -- that is, a way to induce a Dirac structure on a quotient of a submanifold -- respects this relation. After characterizing the minimal scheme of \emph{Dirac reduction}, we prove that two concurring Dirac structures have concurring reductions whenever they share a common \emph{witness}, extending to Dirac geometry the reduction of the Marsden-Ra\cb{t}iu theorem. Two procedures for constructing such common witnesses are given, the second being the Dirac counterpart of Magri's original recipe in bihamiltonian geometry. Examples drawn from Hamiltonian actions, Dirac-Nijenhuis manifolds, and complex Dirac structures conclude the paper and illustrate our methods.

• The paper demonstrates that minimal Dirac reduction generally fails to maintain concurrence and proposes adapted witness subbundles to address this issue.
• It details explicit constructions, including kernel-based methods and a generalization of Magri's recipe, to ensure simultaneous reduction of concurring Dirac structures.
• The study extends classical Marsden–Raţiu reduction techniques to complex settings like Hamiltonian actions, Dirac-Nijenhuis, and complex Dirac structures, preserving geometric compatibility.

Concurring Reduction Schemes for Dirac Structures

Introduction and Context

This paper investigates the interactions between reduction procedures for Dirac structures and the notion of concurrence, a compatibility relation generalizing the commutativity of Poisson structures. Dirac geometry, which subsumes symplectic and Poisson geometry, provides a setting wherein both foliations and (pre-)symplectic forms—as well as more general geometric structures—can be handled in a unified formalism. While reduction techniques such as the Marsden-Weinstein-Meyer theorem and its Poisson/Marsden--Raţiu generalizations are classical, their adaptation to Dirac geometry and their compatibility with multiple structures (notably bihamiltonian or concurrent Dirac structures) is far from trivial.

The authors rigorously elucidate:

• When Dirac reduction commutes with concurrence,
• Under which additional hypotheses strong compatibility is preserved (i.e., both reduced structures remain concurrent),
• How "witness" subbundles mediate this process,
• Which concrete recipes can guarantee the existence of such common witnesses,
• The behavior in various geometric settings (Hamiltonian actions, Dirac-Nijenhuis manifolds, complex Dirac structures).

Dirac Structures, Concurrence, and Reduction

Let $L_M$ and $R_M$ be Dirac structures on $M$. Concurrence is present if their cotangent product $L_M \circledast R_M$ is a smooth Dirac structure; in the Poisson case this reduces to the commutativity of the respective bivectors. The central question addressed is: when a reduction procedure transports a pair of concurring Dirac structures on $M$ to a quotient $Y$, does it preserve their concurrence? For Poisson and symplectic examples, the answer is typically yes, but for general Dirac structures, classical reduction schemes may fail.

A critical technical achievement is the identification of a "minimal" Dirac reduction and the proof that, in general, it does not respect concurrence—a fact supported by explicit counter-examples. The minimal Dirac reduction is characterized by the derived bundle $L_M[I]$ being Dirac along $i$, where $I=F\oplus N^*X$ ties together the relevant vertical and conormal bundles to the substructure.

Witnesses and Concurring Reduction Schemes

To remediate the incompatibility of minimal Dirac reduction, the paper introduces the notion of an adapted witness subbundle: $E \subset TM|_X$. The presence of a common witness for multiple Dirac structures on $R_M$0 ensures that reduction preserves concurrence. Formally, $R_M$1 must satisfy a set of natural compatibility and regularity conditions relative to the structures' interplay and the reduction diagram.

The principal theorem establishes:

If concurring Dirac structures $R_M$2, $R_M$3 admit a common witness $R_M$4, then the reduced structures on $R_M$5 are concurring.

This "concurring reduction" extends to Dirac geometry the mechanism underlying Marsden–Raţiu reduction in the Poisson case and encompasses several classical results as specializations.

Construction of Witnesses

Two explicit general constructions for witnesses are given:

• Reduction by Kernels: The characteristic distributions $R_M$6 and $R_M$7 serve as canonical witnesses (when regular), allowing the construction of reduction diagrams supporting simultaneous reduction of several Dirac structures, all remaining concurring after reduction.
• Magri's Recipe: Extending Magri's classical construction for bihamiltonian structures, for $R_M$8 concurring, the kernel of a derived Dirac structure (combining both the tangent and cotangent product with suitable conjugation) produces a simultaneous witness. This is shown to faithfully generalize the Poisson case and to preserve the essential algebraic structure under reduction.

Geometric Applications and Examples

The methodology is concretely instantiated across several domains:

• Hamiltonian Actions: The classical momentum map scenario is revisited, and the notion of "good values" is translated into the Dirac framework. The orbit distribution associated with group actions is shown to provide witnesses under mild regularity and transversality conditions.
• Dirac-Nijenhuis Manifolds: The theory is shown to be robust under compatible endomorphisms (Nijenhuis structures) enhancing Dirac geometry, with strong functorial properties under reduction.
• Complex Dirac Structures: The formalism is extended to the complex setting, allowing for the manipulation of generalized complex geometry and the extension of Magri-type constructions to complex Poisson structures.

The authors further elucidate the limitations of alternative reduction schemes (e.g., those based on isotropic subbundles or two-form shifts) when it comes to concurrence, documenting explicit counterexamples.

Theoretical Implications

The introduction and axiomatization of the witness-based reduction scheme significantly clarify and unify various reduction scenarios in generalized geometry. It lays a firm foundation for the systematic construction of bihamiltonian (or more generally, multihamiltonian) reduced spaces and associated integrable systems in the Dirac framework. Additionally, it provides a dictionary relating the folklore and frequently invoked "Marsden–Raţiu-type" reductions in the literature to a precise compatibility criterion.

The results suggest that, in the context of integrable systems, bi- or multihamiltonian structures on phase spaces descend, under strong geometric control, to reduced spaces—provided that suitable witness distributions can be constructed. The Magri construction's generalization further supports the extension of separation of variables and spectral curve techniques to the Dirac setting.

Numerical and Structural Claims

While the paper is not computational in nature, it includes several powerful structural assertions:

• The minimal Dirac reduction is incompatible with concurrence in general (contradicting naive expectations).
• For any pair of concurring Dirac structures, their characteristic distributions serve as universal, simultaneous witnesses permitting commute reduction.
• Magri's recipe for bihamiltonian reduction extends to general Dirac structures, provided suitable regularity holds.

Future Directions

The approach outlined has several immediate extensions and open questions:

• Algorithmic construction and classification of witnesses in singular geometries.
• Extension to settings with background 3-forms and higher (Courant) structures.
• Applications to generalized complex, holomorphic Poisson, and higher-Dirac structures.
• Deeper analysis of the consequences for the theory of integrable hierarchical systems.

Conclusion

This work provides a rigorous, unified theoretical structure for the reduction of multiple compatible Dirac structures, culminating in the introduction of witness-mediated concurring reduction schemes. The results have substantial implications for the preservation of algebraic compatibility in reduced systems and pave the way for further advances in the geometric theory of compatible structures and integrable system reduction. The identification and systematization of "witness" subbundles as mediators of compatible reduction constitute a key structural insight with applications across symplectic, Poisson, Dirac, and generalized geometry.