Classification of arithmetic root systems (Q957976): Difference between revisions

A method to classify pointed Hopf algebras was proposed in [\textit{N. Andruskiewitsch} and \textit{H.-J. Schneider}, J. Algebra 209, No. 2, 658--691 (1998; Zbl 0919.16027), and in ``Recent developments in Hopf algebra Theory'', New directions in Hopf algebras. Cambridge: Cambridge University Press. Math. Sci. Res. Inst. Publ. 43, 1--68 (2002; Zbl 1011.16025)]. The method consists in several steps, each step proposing some different problem. The crucial step is the determination of all Nichols algebras satisfying some finiteness condition and associated to braided vector spaces in some class, that depends on the specific class of pointed Hopf algebras one is looking to classify. To the class of finite-dimensional pointed complex Hopf algebras with abelian group, it corresponds the class of braided vector spaces of diagonal type. More precisely, let \((q_{ij})\) be a complex matrix of size \(\theta\) such that \(q_{ij}\) and \(q_{ii}\neq 1\) are roots of 1, for all \(i,j\). Consider a vector space \(V\) with basis \(x_1, \dots, x_\theta\) and \(c\in \text{GL}(V\otimes V)\) given by \(c(x_i\otimes x_j) = q_{ij} x_j\otimes x_i\) for all \(i,j\); then \(c\) is a solution of the braid equation and the concrete problem is to find all matrices \((q_{ij})\) such that the Nichols algebra \(\mathcal B(V)\) is finite-dimensional. For instance, to show the complexity of this problem, let \((a_{ij})\) be a symmetrizable generalized Cartan matrix, let \(D = (d_1, \dots, d_\theta)\) be an invertible diagonal matrix such that \(DA\) is symmetric, and let \(q\) be a root of 1 (with some restrictions on the order). If \(q_{ij} := q^{d_ia_{ij}}\), then \(\mathcal B(V) \simeq \mathfrak u_q^+(\mathfrak g)\), the ``positive'' part of the small quantum group associated by Lusztig to the Kac-Moody algebra \(\mathfrak g\) corresponding to \(A\). Hence \(\mathcal B(V)\) is finite-dimensional iff \(A\) is of finite type. Motivated by this example, the notion of braided vector space of Cartan type was introduced in [\textit{N. Andruskiewitsch} and \textit{H.-J. Schneider}, Adv. Math. 154, No. 1, 1--45 (2000; Zbl 1007.16027)]; this is one given by a matrix \((q_{ij})\) such that \(q_{ii}\neq 1\) and \(q_{ij}q_{ji} = q_{ii}^{a_{ij}}\), for some generalized Cartan matrix \((a_{ij})\). Again, \(\mathcal B(V)\) is finite-dimensional iff \(A\) is of finite type, see \textit{loc. cit.}, under some restrictions on the orders of the \(q_{ij}\)'s. In a parallel development, the existence of a Poincaré-Birkhoff-Witt basis in \(\mathcal B(V)\) for an arbitrary braided vector space \(V\) of diagonal type was established in [\textit{V. Kharchenko}, Algebra Logika 38, No. 4, 476--507 (1999); translation in Algebra Logic 38, No. 4, 259--276 (1999; Zbl 0936.16034)]. By a thorough analysis of the PBW-basis, all matrices \((q_{ij})\) with size \(\theta =2\) such that the Nichols algebra \(\mathcal B(V)\) is finite-dimensional were found in [\textit{I. Heckenberger}, Algebr. Represent. Theory 11, No. 2, 115--132 (2008; Zbl 1175.17003)]. Further, the determination of all braided vector spaces of Cartan type with finite-dimensional Nichols algebra was obtained in [\textit{I. Heckenberger}, Invent. Math. 164, 175--188 (2006; Zbl 1174.17011)], without the restrictions on the orders of the \(q_{ij}\)'s alluded above. In this last paper, a fundamental new tool was introduced, the Weyl groupoid of a braided vector space of diagonal type. The main result in the article under review is the classification of all braided vector spaces of diagonal type with finite-dimensional Nichols algebra. The outcome is several new infinite families of matrices \((q_{ij})\), besides those of Cartan type. The proof is based on the analysis of the PBW-basis and a systematic use of the Weyl groupoid. This Theorem is fundamental for the classification of pointed Hopf algebras with abelian group, but it has also consequences for the study of those with non-abelian group. As the author says, several important open problems remain: to describe the defining relations of these Nichols algebras and to clarify the relationship with Lie superalgebras.