Quantum symmetric algebras (Q5934275)

Let \(V\) be a vector space and let \(c: V \otimes V \to V \otimes V\) be a solution of the braid equation: \((c\otimes \text{id})(\text{id} \otimes c)(c\otimes \text{id}) = (\text{id} \otimes c)(c\otimes \text{id}) (\text{id} \otimes c)\). There is a remarkable braided graded Hopf algebra \(\mathfrak B (V) = \oplus_{n \geq 0} \mathfrak B^n(V)\), which is connected, generated in degree one, with \(\mathfrak B^1(V) \simeq V\) as braided vector spaces, and such that all its primitive elements have degree one. This remarkable braided Hopf algebra was first described by Nichols in his thesis [\textit{W. Nichols}, Commun. Algebra 6, 1521-1552 (1978; Zbl 0408.16007)], and rediscovered in different forms by \textit{S. L. Woronowicz} [Commun. Math. Phys. 122, 125-170 (1989; Zbl 0751.58042)], \textit{G. Lusztig} [Introduction to quantum groups, Prog. Math. 110 (Boston, Birkhäuser) (1993; Zbl 0788.17010)], and \textit{M. Rosso} [C. R. Acad. Sci., Paris, Sér. I 314, No. 1, 5-8 (1992; Zbl 0753.17027)]; see also \textit{J. A. Green} [Shuffle algebras, Lie algebras and quantum groups, Textos de Matemática. Série B. 9. Coimbra: Universidade de Coimbra (1997; Zbl 0919.16029)]. The algebra \(\mathfrak B (V)\) has an alternative description by \(\mathfrak B^n(V) = \text{im }\Omega^{(n)}\), where \(\Omega^{(n)}: T^n(V) \to T^n(V)\) is the so-called ``quantum symmetrizer''. The explicit description of \(\mathfrak B(V)\) is very difficult; it is known that \(\mathfrak B(V) = T(V)\) generically. The authors are concerned with the case when \(c\) is diagonal, that is, when \(V\) has a basis \(x_1, \dots, x_n\) such that \(c(x_i \otimes x_j) = q_{ij} x_j \otimes x_i\) for all \(i\), \(j\), for some non-zero scalars \(q_{ij}\). For this, they work over the field \(\mathbb Q(t_{ij})\) of rational functions on indeterminates \(t_{ij}\), \(i, j\) in \(I\), a set with \(n\) elements. They consider a vector space \(\mathbb V\) over this field with a basis \(X_1, \dots, X_n\) and the ''generic'' braiding \(c(X_i \otimes X_j) = t_{ij} X_j \otimes X_i\) for all \(i\), \(j\). The algebra \(\mathfrak B(\mathbb V)\) has a natural \(\mathbb N^I\)-grading; if \(\alpha\) in \(\mathbb N^I\) has length \(n\), then \(\mathfrak B(\mathbb V)^{\alpha}\) is stable under \(\Omega^{(n)}\). Let \(\Omega^{\alpha}\) be the restriction of \(\Omega^{(n)}\) to \(\mathfrak B(\mathbb V)^{\alpha}\). The main result of this paper says: Any irreducible factor of \(\det \Omega^{\alpha}\) is a divisor of \(1 - u(\beta)\), for some \(\beta \leq \alpha\), length of \(\beta \geq 2\). Here \(u(\beta)\) is defined as follows: If \(X_{i_1} \dots X_{i_M}\) is any monomial in \(\mathfrak B(\mathbb V)^{\beta}\) then \(u(\beta) = \prod_{1 \leq h \neq j \leq M} t_{i_h, i_j}\). Furthermore, they describe \(\det \Omega^{\alpha}\) as a product of irreducible factors when \(\alpha\) is multiplicity-free. Reviewer's remark. An important problem about the algebras \(\mathfrak B(V)\) is to determine when they have finite dimension, or finite Gelfand-Kirillov dimension. Answers to these questions are given in [\textit{M. Rosso}, Invent. Math. 133, No. 2, 399-416 (1998; Zbl 0912.17005)] (in the case when the \(q_{ij}\)'s are positive) and [\textit{N. Andruskiewitsch} and \textit{H.-J. Schneider}, Adv. Math. 154, No. 1, 1-45 (2000; Zbl 1007.16027)] (in the case when the \(q_{ij}\)'s are roots of 1 and \(c\) is of Cartan type). In this last paper, the authors call \(\mathfrak B(V)\) a ``Nichols algebra'' since in general these algebras seem to be far from symmetric algebras, \textit{e.~g.} they are tensor algebras generically. Examples of finite dimensional Nichols algebras where \(c\) is not of Cartan type exist, see for instance [W. Nichols, \textit{loc. cit.}]. The results in the present paper may be useful to study further this subclass of Nichols algebras. For related material, see also \textit{V. K. Kharchenko} [J. Algebra 217, No. 1, 188-228 (1999; Zbl 0943.17016) and 238, No. 2, 534-559 (2001; Zbl 0987.16031)].