Combinatorics and invariant differential operators on multiplicity free spaces. (Q1868751): Difference between revisions

Let \(U\) be a finite dimensional complex vector space. Then the algebra \(PD(U)\) of linear differential operators with polynomial coefficients has the following presentation: it is generated by \(U\) (the directional derivatives) and \(U^*\) (the linear functions) which satisfy the following relations: \[ [\partial_1,\partial_2]=0,\quad[x_1,x_2]=0,\quad [\partial,x]=\partial(x) \] for all \(\partial_1,\partial_2,\partial\in U\), \(x_1,x_2,x\in U^*\). This implies that there is a unique antiautomorphism \(D\to{^tD}\) of \(PD(U)\) with \[ ^t\partial=-\partial,\quad ^tx=x,\quad\partial\in U,\;x\in U^*. \] Let \(G\) be an algebraic group acting linearly on \(U\). Then transposition is \(G\)-equivariant. It induces an antiautomorphism of the algebra \(PD(U)^G\) of \(G\)-invariant differential operators. Assume that \(G\) is connected, reductive and \(U\) a multiplicity free space. By assumption, the algebra \(P(U)\) decomposes as a \(G\)-module as \(P(U)=\bigoplus_{\lambda\in\Lambda_+}P^\lambda\). Every \(D\in PD(U)^G\) acts on \(P^\lambda\) as multiplication by a scalar, which is denoted by \(c_D(\lambda)\). Using this weight, the author defines a new embedding \(PD(U)^G\hookrightarrow P(V)\): \(D\mapsto p_D\), where \(p_D(v):=c_D(v-\rho)\). Theorem 2.2. Let \(U\) be a multiplicity free space for \(G\). Then \(p_{^tD}(v)=p_D(-v)\), \(D\in PD(U)^G\), \(v\in V\). For every \(h\in P^W\), the author defines the difference operator \(D_h:=\exp(\text{ad\,}L)(h)\). Theorem 3.5. Let \(h\in P^W\). Then \(D_h(p_\lambda)=h(\rho+\lambda)p_\lambda\), for all \(\lambda\in\Lambda_+\). Corollary 4.6. Let \(\rho\in V_0\) be strongly dominant. Then for all \(\mu\in\Lambda_+\) holds \(p_\mu(-\rho)=(-1)^{l(\mu)}d_\mu\).