Criteria for the finiteness of restriction of \(U({\mathfrak g})\)-modules to subalgebras and applications to Harish-Chandra modules (Q2366788)

Let \({\mathfrak g}\) be a finite-dimensional complex Lie algebra, \(U({\mathfrak g})\) its enveloping algebra, \(S({\mathfrak g})_ +\) the augmentation ideal of the symmetric algebra \(S({\mathfrak g})\). In this paper the author gives a criterion for a finitely generated \(U({\mathfrak g})\)-module \(H\) to stay finitely generated on restriction to a subalgebra \(A\) of \(U({\mathfrak g})\), in terms of the associated variety \({\mathcal V}\subset{\mathfrak g}^*\) of \(H\). More precisely, assume that the graded algebra \(\text{gr }A\subset\text{gr }U({\mathfrak g})\cong S({\mathfrak g})\) of \(A\) is Noetherian. Then the author asserts that the associated graded module of \(H\) is finitely generated over \(\text{gr }A\) if and only if \(\mathcal V\) does not meet the associated variety of the ideal generated by \(\text{gr }A\cap S({\mathfrak g})_ +\) in \(S({\mathfrak g})\). In case \(H\) satisfies this condition and \(A\) happens to be the enveloping algebra of a Lie subalgebra \({\mathfrak n}\) of \({\mathfrak g}\), then the notions of Gelfand- Kirillov dimension and multiplicity coincide over \({\mathfrak g}\) and \({\mathfrak n}\). If in addition \({\mathfrak g}\) is real semisimple, then it is often the case that the above condition is satisfied by any Harish-Chandra module \(H\). In particular this holds when \({\mathfrak n}\) is a symmetrizing subalgebra, a maximal nilpotent Iwasawa subalgebra, or a quasi-spherical subalgebra. In these cases one gets a bound for the dimension of the space of intertwining operators between a Harish-Chandra module and a \(U({\mathfrak g})\)-module induced from a \(U({\mathfrak n})\)-module. This bound appeared earlier in [\textit{H. Yamashita}, J. Math. Kyoto Univ. 28, 173-211 (1988; Zbl 0734.22004)] in a special case. No proofs are given here; they will appear in a later paper.