Integrability of quotients in Poisson and Dirac geometry (Q2057569)

This paper studies the integrability of Poisson structures (and, more in general, Lie algebroids) that arise from quotient constructions. In this sense it generalises and unifies some previous integrability results that appeared in [\textit{H. Bursztyn} et al., ``Dirac geometry and integration of Poisson homogeneous spaces'', Preprint, \url{arxiv:1905.11453}; \textit{R. L. Fernandes} and \textit{D. Iglesias Ponte}, Lett. Math. Phys. 90, No. 1--3, 137--159, (2009; Zbl 1183.53075)]. Indeed, a possible approach to studying the integrability of a Poisson structure arising from reduction of a Dirac structure consists in first understanding the integrability of the original Dirac structure and then checking whether one of its integrations gives rise by reduction to an integration of the quotient Poisson structure. First, the paper addresses the more general question: \textit{``Given a surjective submersion \(q\colon S\to M\) and a Lie algebroid \(A\to M\), how is the integrability of \(A\) related to the integrability of the pullback Lie algebroid \(q^!A\to S\)?''} Clearly, the integrability of \(A\) implies the integrability of \(q^!A\) (cf. Corollary 1.9 in [\textit{P. J. Higgins} and \textit{K. Mackenzie}, J. Algebra 129, No. 1, 194--230 (1990; Zbl 0696.22007)]) but the converse does not hold (Example 3.8). Theorem 3.2 answers the previous question as follows: \(A\) is integrable if and only if \(q^!A\) admits a \textit{\(q\)-admissible integration}, i.e., \begin{itemize} \item the pullback Lie algebroid \(q^!A\Rightarrow S\) is integrable and \item the inclusion \(\ker Tq\to{}q^!A\) is integrable by a Lie groupoid morphism \(\Phi\colon S\times_MS\rightarrow G\), where \(S\times_MS\rightrightarrows S\) is the submersion groupoid. \end{itemize} Moreover, if \(q^!A\) admits a \(q\)-admissible integration as above, and \(R\) denotes the equivalence relation on \(G\) given, for all compatible \(g\in G\) and \(x,y\in S\times_MS\), by \[ g\sim\Phi(x)g\Phi(y), \] then Theorem 3.2 proves that the quotient \(G/R\) inherits a unique Lie groupoid structure over \(M\) such that the quotient map \(Q\colon G\rightarrow G/R\) is a Lie groupoid morphism. Additionally, the quotient Lie groupoid \(G/R\) integrates \(A\), i.e. \(\operatorname{Lie}(G/R)\cong A\). The main result of the paper, Theorem 3.10, specializes Theorem 3.2 to the case of Poisson and Dirac structures. So, given a surjective submersion \(q\colon S\to M\), it characterizes the integrability of a Poisson structure \(\pi\) on \(M\) in terms of the pullback Dirac structure \(q^!(\operatorname{graph}(\pi))\) on \(S\). Specifically, it proves that the Poisson structure \(\pi\) is integrable if and only if the Dirac structure \(q^!(\operatorname{graph}(\pi))\) admits a \emph{\(q\)-admissible presymplectic integration} \((G,\omega)\), i.e. \begin{itemize} \item the Dirac structure \(q^!(\operatorname{graph}(\pi))\) structure admits a \(q\)-admissible integration \(G\), with associated Lie groupoid morphism \(\Phi\colon S\times_MS\to G\), and \item \(G\) is endowed with a multiplicative presymplectic form \(\omega\) such that its infinitesimal counterpart is provided by \(q^!(\operatorname{graph}(\pi))\) and \(\Phi^\ast\omega=0\), \end{itemize} Moreover, Theorem 3.10 also proves that there is a (unique) multiplicative symplectic form \(\overline{\omega}\in\Omega^2(G/R)\) such that \(Q^\ast\overline{\omega}=\omega\). As an immediate consequence of its main result the paper recovers well-known results about coisotropic reduction (Proposition 3.13) and the weak Morita invariance of integrations (Proposition 3.15). Additionally, the paper also applies Theorem 3.10 to get entirely new results like the integrability of two special classes of geometric structures, namely: \begin{itemize} \item Dirac homogeneous spaces (Theorem 4.19) and \item Poisson homogeneous spaces of symplectic groupoids integrationg Poisson groups (Theorem 4.36). \end{itemize}