Bi-isotropic decompositions of semisimple Lie algebras and homogeneous bi-Lagrangian manifolds (Q2370293): Difference between revisions

In the first part of this paper the authors investigate bi-isotropic decompositions of a semisimple Lie algebra, and in the sequel they provide two interesting geometrical applications. The first is related to certain decompositions of the tangent bundle of a homogeneous space \(M=G/K\), and the second to the classification of invariant bi-Lagrangian (or equivalently para-Kähler) structures on \(M\). More precisely: Let \(\mathfrak g\) be a semisimple Lie algebra (over \(\mathbb R\) or \(\mathbb C\)) with Killing form \(B\), and \(\mathfrak k\subset\mathfrak g\) a \(B\)-nondegenerate subalgebra of \(\mathfrak g\). A decomposition \( \mathfrak g=\mathfrak k +\mathfrak m^+ +\mathfrak m^- \) into \(\text{ad}_{\mathfrak k}\)-submodules is called a bi-isotropic decomposition if (i) the subspaces \(\mathfrak m^{\pm}\) are \(B\)-isotropic (i.e. \(B| _{\mathfrak m^{\pm}}=0\)), (ii) the subspace \(\mathfrak m:= \mathfrak m^+ +\mathfrak m^-\) is \(B\)-orthogonal, and (iii) \(\mathfrak g^{\pm}:=\mathfrak k+\mathfrak m^{\pm}\) are subalgebras. The authors give a description of all bi-isotropic decompositions of \(\mathfrak g\) with \(\mathfrak k\) a subalgebra of maximal rank. It is known that in this case the centralizer \(Z_{\mathfrak g}(h)\) of a semi-simple element \(h\in\mathfrak g\), and a bi-isotropic decomposition is a generalized Gauss decomposition. The description of such decompositions for a given \((\mathfrak g, \mathfrak k)\) is reduced to a description of all parabolic subalgebras \(\mathfrak g^+\) with given reductive part \(\mathfrak k\). This problem was solved some time ago by [\textit{D. V. Alekseevsky} and \textit{A. M. Perelomov}, Funct. Anal. Appl. 20, 171--182 (1986); translation from Funkts. Anal. Prilozh. 20, No. 3, 1--16 (1986; Zbl 0641.53050)]. At the present paper it is solved for the real case in terms of the crossed Satake diagrams.