Reducibility of generalized Verma modules for Hermitian symmetric pairs (Q2207271)

1. Notations and definitions in the present paper. Let \(G\) be a simple real Lie group with a maximal compact subgroup \(K\) such that \(G/K\) is a Hermitian symmetric space. Let \(\mathfrak{g}_{0}\) and \(\mathfrak{k}_{0}\) be the corresponding Lie algebras with a Cartan decomposition \(\mathfrak{g}_{0}=\mathfrak{k}_{0}\oplus \mathfrak{p}_{0}\). The classification of the pair \((\mathfrak{g}_{0},\mathfrak{k}_{0})\) and corresponding root systems are well-known (see Table 1). We drop the subscript \(0\) to represent complexifications of Lie algebras. If \(\mathfrak{h}\) is a Cartan subalgebra of \(\mathfrak{k}\), then it is also a Cartan subalgebra of \(\mathfrak{g}\). The root system \(\Delta\) can be viewed as a subset of a real space \(\mathbb{R}^{n}\) with standard basis \{\(e_{1},\cdot \cdot \cdot, e_{n}\)\}, that is, \(<e_{i}, e_{j}>=\delta_{ij}\) for \(1 \le i, j \le n\). Let \(\Delta \) be the root system of \((\mathfrak{g},\mathfrak{h})\), and let \(\Delta _{c}\) be the root system of \((\mathfrak{k},\mathfrak{h})\). Then \(\Delta_{c}\) is called the set of compact roots and \(\Delta_{nc}:=\Delta \backslash\Delta _{c}\) is called the set of noncompact roots. Denote by \(W\) (resp. \(W_{c}\)) the Weyl group generated by reflections \(s_{\alpha }\) with \(\alpha \in\Delta \) (resp. \(\alpha \in\Delta _{c}\)). Let \(l(w)\) be the length function on \(W\). The action of the reflection \(s_{\alpha }\) on \(\lambda \in \mathfrak{h}^{*}\) is defined by \(s_{\alpha }\lambda =\lambda -<\lambda,\alpha ^{\vee}>\alpha .\) Here \(<\cdot\), \(\cdot>\) denotes the bilinear form on \(\mathfrak{h}^{*}\) induced from the Killing form of \(\mathfrak{g}\) and \(\alpha ^{\vee}:=2\alpha /<\alpha ,\alpha >\) is the coroot of \(\alpha \). Choose a positive system \(\Delta ^{+}\) of \(\Delta \), so that it determines a decomposition \(\mathfrak{p}=\mathfrak{p}^{+}\oplus \mathfrak{p}^{-}\), where \(\mathfrak{p}^{+}\) and \(\mathfrak{p}^{-}\) are abelian. Then \(\mathfrak{q}=\mathfrak{k}\oplus \mathfrak{p}^{+}\) is a parabolic subalgebra of \(\mathfrak{g}\). Write \(\Delta _{c}^{+}:=\Delta _{c}\cap\Delta ^{+}\) and \(\Delta _{nc} ^{+}:=\Delta _{nc}\cap\Delta ^{+}\). Set \(\Lambda _{c}^{+}:=\{\lambda \in h^{\ast}|<\lambda ,\alpha ^{\vee}>\in \mathbb{Z}_{>0}\) for all \(\alpha \in\Delta _{c}^{+}\}.\) For \(\lambda \in\Lambda _{c}^{+}\) , let \(F(\lambda )\) be the finite-dimensional irreducible \(\mathfrak{k}\)-module of highest weight \(\lambda -\rho \), where \(\rho \) is half the sum of roots in \(\Delta ^{+}\). By letting \(\mathfrak{p}^{+}\) act trivially on \(F(\lambda )\), we may consider \(F(\lambda )\) as a module of \(\mathfrak{q}\). The generalized Verma module \(N(\lambda )\) is defined to be \(N(\lambda ):=U(\mathfrak{g})\otimes_{U(\mathfrak{q})}F(\lambda )\). 2. \textit{J. C. Jantzen}'s reducibility criterion: Theorem 3.3 ([Math. Ann. 226, 53--65 (1977; Zbl 0372.17003), Satz 3]) Usually we use Jantzen's criterion [loc. cit.] to determine the reducibility of generalized Verma modules. Let \(e^{\lambda }\) be the \(\mathbb{Z}\)-valued function on \(\mathfrak{h}^{*}\) which takes the value \(1\) at \(\lambda\) and \(0\) elsewhere. Let \(\lambda \in\Lambda _{c}^{+}\). The \(\mathfrak{g}\)-module \(N(\lambda )\) is irreducible if and only if \(\sum\limits_{\beta \in\Delta _{\lambda }^{+}}Y(s_{\beta }\lambda )=0\), where \(Y(\lambda )=D^{-1}\sum\limits_{w\in W}c(-1)^{l(w)}e^{w\lambda }\) , \(D=e^{\rho }\prod\limits_{\alpha \in\Delta ^{+}}(1-e^{-\alpha })\) is the Weyl denominator. As described above, Jantzen's criterion is not easy to apply in general. 3. The main result in the present paper: Theorem 5.1, 5.2, 5.3, 5.4, 5.6, 5.7, and Theorem 5.10. The authors give explicit reducibility criteria for generalized Verma modules associated with each Hermitian symmetric pair. For \(\lambda=\lambda_{1}e_{1}+\lambda_{2}e_{2}+\cdot\cdot\cdot+\lambda_{n}e_{n}\) \(\in\Lambda _{c}^{+}\), the generalized Verma module \(N(\lambda )\) is reducible if and only if we can find \(\lambda\)'s indices \(i_{0}\), \(j_{0}\) which satisfy the certain conditions. The author's criterion are given by the information of the weight's indices and are relatively easy to verify. Example 1.1. Let \(\mathfrak{g} = \mathfrak{su}(3, 4)\). Consider \(\lambda = (8, 5, -13, 8, 3, 2, -13) \in \Lambda^{+}_{c}\). Then \(S_{\lambda} = \{ \pm (e_{1}-e_{4}), \pm (e_{3}-e_{7})\}\) and \((S_{\lambda}^{+})^{\perp} = \{ \pm(e_{2}-e_{5}), \pm(e_{2}-e_{6}), \pm(e_{5}-e_{6})\}\). \(i_{0}\) is the smallest index in \(\{1, 2, 3\}\) for which \(\lambda_{i_{0} }- \lambda_{j} \neq 0\) whenever \(3+1 \le j \le 3+4\), and \(j_{0}\) is the largest index in \(\{3+1, \cdot \cdot \cdot, 3+4\}\) for which \(\lambda_{i}- \lambda_{j_{0}} \neq 0\) whenever \(1 \le i \le 3\). So \(\beta_{\lambda} = e_{2} - e_{6}\) . With \(\lambda_{2}- \lambda_{6} = 3 \in \mathbb{Z}_{>0} (i_{0}=2,j_{0}=6)\), the module \(N(\lambda)\) is reducible. \(\underline{\text{Table 1: Hermitian symmetric pairs. }}\) \(\underline{\text{ }\mathfrak{g}_{0}\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \mathfrak{k}_{0}\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }\) \(\text{ }\mathfrak{su}(p,q)\ \ \ \ \ \ \ \ \ \ \ \ \ \mathfrak{su}(p)\oplus \mathfrak{su}(q)\oplus \mathbb{R} \ \ \) \(\text{ }\mathfrak{sp}(n,\mathbb{R}) \ \ \ \ \ \ \ \ \ \ \ \mathfrak{su}(n)\oplus \mathbb{R} \ \ \ \ \ \ \ \ \ \ \ \ \) \(\text{ }\mathfrak{so}^{\ast}(2n) \ \ \ \ \ \ \ \ \ \ \ \ \mathfrak{su}(n)\oplus \mathbb{R} \ \ \ \ \ \ \ \ \ \ \ \ \ \) \(\text{ }\mathfrak{so}(2,2n-2) \ \ \ \ \ \mathfrak{so}(2n-2)\oplus \mathbb{R} \ \ \ \ \ \ \ \) \(\text{ }\mathfrak{so}(2,2n-1) \ \ \ \ \ \mathfrak{so}(2n-1)\oplus \mathbb{R} \ \ \ \ \ \ \ \) \(\text{ }EIII \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \mathfrak{so}(10)\oplus \mathbb{R} \ \ \ \ \ \ \ \ \ \ \ \) \(\underline{\text{ }EVII \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \mathfrak{e}_{6}\oplus \mathbb{R} \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }\)