Generalized double affine Hecke algebras of rank 1 and quantized del Pezzo surfaces (Q884439)

Double affine Hecke algebras for reduced root systems were introduced by \textit{I. Cherednik} [Int. Math. Res. Not. 1992, No. 9, 171--180 (1992; Zbl 0770.17004)] in order to prove the Macdonald conjectures. Those of type \(C^\vee C_n\) were introduced in the works of Noumi, Sahi and Stokman as a generalization of Cherednik algebras of types \(B_n\) and \(C_n\), in order to prove the Macdonald conjectures for Koornwinder polynomials. This article defines and studies new algebras \(H(t,q)\) which are generalizations of double affine Hecke algebras of type \(C^\vee C_n\) in the case \(n=1\): Fix a star-shaped simply laced affine Dynkin diagram \(\widehat{D}\) (\(\tilde{D}_4\),\(\tilde{E}_6\),\(\tilde{E}_7\), or \(\tilde{E}_8\)). Let \(m\) be the number of legs of \(\widehat{D}\) and \(d_j-1\), \(j=1,\dots ,m,\) be the length of the \(j\)th leg. Then the authors define a family of algebras \(H(t,q)\) depending on parameters \(q\in\mathbb{C}^\ast\) and \(t=(t_{ij})\), \(t_{kj}\in\mathbb{C}^\ast\), \(k=1,\dots,m,\) \(j=1,\dots,d_k\), by generators \(T_k\), \(k=1,\dots,m\), with explicitly defined relations \[ \prod_{j=1}^{d_k}(T_k-e^{2\pi ij/d_k}t_{kj})=0,\text{ }k=1,\dots,m;\text{ }\prod_{k=1}^m T_k=q. \] For \(\widehat{D}=\tilde{D}_4\) the result is exactly the double affine Hecke algebra of type \(C^\vee C_1.\) In the cases \(\widehat{D}=\tilde{E}_{6,7,8}\) the results are new algebras which are the main subject of this article. If \(t_{kj}=1\) and \(q=1\), the algebra \(H(t,q)=H(1,1)\) is the group algebra for some group \(G\) known to be isomorphic to a 2-dimensional crystallographic group \(\mathbb{Z}_l\ltimes\mathbb{Z}^2,\) \(l=2,3,4,6\) in the cases \(\widehat{D}=\tilde{D}_4,\tilde{E}_6,\tilde{E}_7,\tilde{E}_8,\) respectively. Moreover, \(H(1,q)\) is a twisted group algebra of \(G\) so that \(H(t,q)\) is a deformation of the twisted group algebra of \(G\). The authors prove that if they regard \(\log(t_{kj})\) as formal parameters then this deformation is flat, and \(H(t,qe^{\varepsilon})\) is the universal deformation of \(H(1,q)\) if \(q\) is not a root of unity (\(\varepsilon\) is a new formal parameter). The authors also prove a more delicate algebraic PBW-theorem, claiming that some filtrations on \(H(t,q)\) have certain explicit Poincaré series, independent of \(t\) and \(q\). It is known that for \(\widehat{D}=\tilde{D}_4\) and \(q=1\) the algebra \(H(t,q)\) is finite over its center \(Z(t,q)\), and the spectrum of \(Z(t,q)\) is an affine cubic surface, obtained from a projective one by removing three lines forming a triangle. The authors prove that this result is valid also for \(q\) being a root of unity, and they generalize it to the cases \(\widehat{D}=\tilde{E}_{6,7,8}\). In these cases the spectrum of \(Z(t,q)\), \(q\) a root of unity, turns out to be an affine surface \(S(t,q)\) obtained from a projective del Pezzo surface \(S(t,q)\) of degrees \(3,2,1\) respectively by removing a nodal \(\mathbb{P}^1.\) This means that for \(q\neq 1\), the spherical subalgebra \(eH(t,q)e\) in \(H(t,q)\) should be viewed as an algebraic quantization of the surface \(S(t,1)\). Moreover, the algebraic PBW theorem for \(H(t,q)\) implies that the Rees algebra of \(eH(t,q)e\) with respect to an appropriate filtration provides a (noncommutative) quantization of the Poisson surface \(S(t,q)\). The authors recall the basics of crystallographic groups in the plane. They consider the twisted group algebra \(B(q)=H(1,q)\) of a planar crystallographic group \(G\), and deform it into an algebra \(\mathbf{H}(q)\), which is a version of \(H(t,q)\) in which \(\log t_{kj}\) are formal parameters. They prove the formal PBW theorem for \(\widehat{\mathbf{H}}(q)\), and give results on the cohomology of \(B(q)\) and on its universal deformation. The authors then define an increasing filtration on \(H(t,q)\), the length filtration. They prove that the Poincaré series for this filtration is independent of \(t,q\). They then use this result to establish general properties of \(H(t,q)\). Using the Riemann-Hilbert correspondence they define a homomorphism from a formal version of the generalized double affine Hecke algebra to the completion of the deformed preprojective algebra of the quiver associated to the graph \(\widehat{D}\). Thus a holomorphic map from the universal deformation of the Kleinian singularity \(\mathbb{C}^2/\Gamma\) to the family of surfaces \(S(t,1)\) can be defined. This is a local isomorphism of analytic varieties near \(0\in\mathbb{C}^2/\Gamma\), which in the case \(\widehat{D}=\tilde{D}_4\) encodes generic solution to the Painlevé VI equation. The authors study closely the surfaces \(S(t,q)\): Let \(\mathbf{G}\) be the simple Lie group corresponding to the diagram \(D\). They then prove that the algebra \(H(t,q)\) depends only on the projection of \(t\) to the maximal torus \(\mathbf{T}\subset\mathbf{G}\), and that the map \(t\mapsto S(t,1)\) from \(\mathbf{T}\) to the moduli space of affine del Pezzo surfaces is Galois and has Galois group isomorphic to the Weyl group \(W\) of \(G\). This also implies that the coefficients of the equation of \(S(t,1)\), as functions of \(t\), are polynomials of characters of irreducible representations of \(\mathbf{G}\), and the authors compute these polynomials explicitly. The authors give the necessary definitions and results of deformation theory, and they organize the article such that the results depending on computer calculations are left to the end of the article. This article covers a lot of results depending heavily on nontrivial theories, both analytic and algebraic. It is a very nice article covering nearly everything possible on the algebras \(H(t,q)\) and their deformation theory. It contains high quality results, and it takes some time to fully understand them and their proofs.