Obstructions to the integrability of \(\mathcal{VB}\)-algebroids (Q722387)

Lie groupoids and Lie algebroids are found throughout differential geometry in many guises and there are intimate links with Poisson/symplectic geometry. The Lie theory for Lie algebroids and Lie groupoids has been extensively studied. Unlike the Lie theory of Lie groups and Lie algebras, one cannot always integrate a Lie algebroid to obtain a Lie groupoid, i.e., there is no Lie's third theorem for Lie algebroids. \textit{M. Crainic} and \textit{R. L. Fernandes} [Ann. Math. (2) 157, No. 2, 575--620 (2003; Zbl 1037.22003)] showed that the integrability is controlled by a computable obstruction given in terms of the so-called monodromy groups associated with the Lie algebroid. The paper under review studies the question of integrability of \(\mathcal{VB}\)-algebroids to \(\mathcal{VB}\)-groupoids. That is, they consider vector bundles in the category of Lie algebroids and Lie groupoids. A \(\mathcal{VB}\)-algebroid can be understood as a double structure \[ \begin{tikzcd} D \arrow[r, "p"] \arrow[d, "p_D" left] & A \arrow[d, "p_A"] \\E \arrow[r, "p_E"] & M\end{tikzcd} \] where the vertical arrows are Lie algebroid structures and the horizontal arrows are vector bundle structures. The natural compatibility is that the vertical structure maps must define morphisms for the horizontal structures, and vice-versa. Similar diagrams describe \(\mathcal{VB}\)-groupoids. A theorem of \textit{H. Bursztyn} et al. [Adv. Math. 290, 163--207 (2016; Zbl 1391.53092)] states that a \(\mathcal{VB}\)-algebroid is integrable if and only if the Lie algebroid \(D \rightarrow E\) is integrable. The integrability of \(A \rightarrow M\) naturally follows. (A generalisation of these results was given in [\textit{A. J. Bruce} et al., SIGMA, Symmetry Integrability Geom. Methods Appl. 11, Paper 090, 25 p. (2015; Zbl 1331.22006)]). The main theorem of the paper under review is that a \(\mathcal{VB}\)-algebroid is integrable if and only if \(A \rightarrow M\) is integrable and the spherical periods of a certain cohomology class associated with the \(\mathcal{VB}\)-algebroid vanish. The authors then use this theorem to study the integrability question for \(2\)-term representations up to homotopy which is tightly linked to \(\mathcal{VB}\)-algebroids and \(\mathcal{VB}\)-groupoids via the work of \textit{A. Gracia-Saz} and \textit{R. A. Mehta} [Adv. Math. 223, No. 4, 1236-1275 (2010; Zbl 1183.22002); J. Symplectic Geom. 15, No. 3, 741-783 (2017; Zbl 1387.18009)].