Differential operators and Cherednik algebras (Q2269868)

Let \(G\) be a reductive algebraic group acting on a smooth affine algebraic variety \(X\). This gives rise to a locally finite \(G\)-action on the coordinate ring \(\mathbb{C}[X]\) and on the ring of regular differential operators \(\mathcal{D}(X)\) via the formulas \[ (g\cdot f)(x):=f(g^{-1}\cdot x)\text{ and }(g\cdot\theta)(f):=g\cdot(\theta(g^{-1}\cdot f)) \] for \(g\in G\), \(f\in\mathbb{C}[X]\), \(\theta\in\mathcal{D}(X)\) and \(x\in X\). The action of \(G\) on \(\mathbb{C}[X]\) and \(\mathcal{D}(X)\) gives rise to a \(\mathfrak g\)-action on \(\mathbb{C}[X]\) by derivations, and this induces a Lie algebra map \(\tau:\mathfrak g\rightarrow\text{Der}(\mathbb{C})\subset\mathcal{D}(X)\). This extends uniquely to an associative algebra morphism \(\tau:U(\mathfrak g)\rightarrow\mathcal{D}(X)\). Let \(\nu:\mathfrak g\rightarrow\mathbb{C}\) be a Lie algebra character and write \(I_\nu\) for the two-sided ideal in \(U(\mathfrak g)\) generated by the elements \(\{x-\nu(x):x\in\mathfrak g\}\). The space of \(G\)-invariants \[ [\mathcal{D}(X)/\mathcal{D}(X)\tau(I_\nu)]^G \] with its induced algebraic structure is then called the quantum hamiltonian reduction of \(\mathcal{D}(X)\) at \(\nu\). Let \(W=S_n\), \(n\geq 2\), be the symmetric group and write \(e=|W|^{-1}\sum_{w\in W}w\in\mathbb{C}W\) for the trivial idempotent, \(e_-=|W|^{-1}\sum_{w\in W}\text{sign}(w)\cdot w\in\mathbb{C}W\) denote the sign idempotent. Then \(\mathfrak h=\mathbb C^n\) is the permutation representation of \(W\) and \[ \mathfrak h^{\text{reg}}=\{(z_1,\dots,z_n)\in\mathfrak h:z_i\neq z_j\text{ for all }1\leq i<j\leq n\} \] denotes the subset of \(\mathfrak h\) on which \(W\) acts freely. There is an induced action of \(W\) on \(\mathcal{D}(\mathfrak h^{\text{reg}})\) and \(\mathcal{D}(\mathfrak h^{\text{reg}})\ast W\) denotes the corresponding skew group ring. For a fixed scalar \(c\in\mathbb{C}\), \(\text{H}_c\) denotes the \textit{rational Cherednik algebra} corresponding to \(\mathrm{GL}_n(\mathbb{C})\). \(\text{H}_c\) is identified with the subalgebra of \(\mathcal{D}(\mathfrak h^{\text{reg}})\ast W\) generated by \(W\), the vector space \(\mathfrak h^\ast=\sum x_i\mathbb{C}\subset\mathbb{C}[\mathfrak h]\) of linear functions, and the \textit{Dunkl operators} \[ D_c(y_i)=\frac{\partial}{\partial x_i}-\frac12\sum_{j\neq k}c\frac{\langle y_i,x_j-x_k\rangle}{x_j-x_k}(1-s_{jk}) \] where \(\{y_1,\dots,y_n\}\subset\mathfrak h\) is the dual basis to \(\{x_1,\dots,x_n\}\) and the \(s_{jk}\in W\) are simple transpositions. There is a PBW isomorphism \(\text{H}_c\cong\mathbb{C}[\mathfrak h]\otimes\mathbb{C}W\otimes\mathbb{C}[\mathfrak h^\ast]\) of \(\mathbb{C}\)-vector spaces. The \textit{spherical subalgebra} of \(\text{H}_c\) is \(\text{U}_c=e\text{H}e\). Let \(\chi_c:\mathfrak g\rightarrow\mathbb{C}\) be the Lie algebra homomorphism defined by \(x\mapsto c\cdot\text{tr}(x)\). Letting \(I_c=I_{\chi_c}\subset U(\mathfrak g)\) for short, \(I_c=U(\mathfrak g)\mathfrak{sl}_n+U(\mathfrak g)(1-nc)\). Essential through the article are the objects \(\mathfrak{D}_c:=\frac{\mathcal{D}(\mathfrak{G})}{\mathcal{D}(\mathfrak{G})\tau(I_c)}\) and \(\mathfrak{D}^G_c:=\left[\frac{\mathcal{D}(\mathfrak{G})}{\mathcal{D}(\mathfrak{G})\tau(I_c)}\right]^G\) given the differential operator filtration \(\Gamma\) induced from that on \(\mathcal{D}(\mathfrak{G})\). Cherednik algebras are closely related to the Hilbert scheme \(\text{Hilb}^{n}\mathbb C^{2}\) of points in the plane. Important classes of representations of the Cherednik algebra corresponds to important classes of sheaves on the Hilbert scheme and, through them, to important combinatorial objects. Set \(\mathcal{C}=\{c\in\mathbb C:c=a/b\text{ where }a,b\in\mathbb Z\text{ with }2\leq b\leq n\}\). \(c\in\mathbb C\) is called good provided \(c\notin\mathcal C\cap(-1,0)\). The good values for \(c\in\mathbb C\) are the ones for which \(\text{U}_c\) has good properties. A known theorem states the following: Let \(c-1,\dots,c-m+1, c\in\mathbb C, m\in\mathbb Z\) be good. (1) If \(\Gamma\) denotes the differential operator filtration then one has \(\text{gr}_\Gamma(_{c-m}\text{Q}_c)=\delta^{-m}A^me\) inside \(\text{gr}_\Gamma\text{U}^{\text{reg}}=\mathbb C[\mathfrak h^{\text{reg}}\times\mathfrak h]e.\) (2) Similarly \(\text{gr}_\Gamma(_cP_{c-m})=A^m\delta^me\). This article includes two main aims. First, the authors would like to obtain a proof of the above result that is independent of Haiman's work. One will then be able to apply Cherednik algebra techniques to the study of \(\text{Hilb}^n\mathbb C^2\) and related combinatorial objects. Second, through the noncommutative geometry approach, and through quantum hamiltonian reduction of the \(GL_n(\mathbb C)\)-equivariant space \(\mathfrak{gl}(\mathbb C)\times\mathbb C^n\) one would like to understand the relationship between these two approaches.The authors solve both problems by using quantum hamiltonian reduction to give an alternative and shorter proof of Haiman's result, clarifying the relationship between the two approaches. For example, the authors prove that two different definitions of characteristic cycles on the Hilbert scheme coincide. This means that the second main result in the article is the following: Fix \(c\in\mathbb C\) and an integer \(m\geq 1\) so that each of \(c-1,c-2,\dots,c-m+1\) is good. Under the differential operator filtration on the two sides there is a filtered \((\text{U}_{cm},\text{U}_c)\)-bimodule isomorphism \[ \Theta_{c,m}:\mathfrak D^{\det^{-m}}_{c+1}\overset\sim \rightarrow_{c-m}\text{Q}_c. \] The article is highly technical and not self contained. Besides that, the article is very explicit, logical, and it has deep and interesting results relating two different approaches, in some sense generalizing the theory of Lie groups.