Prime decomposition of modular tensor categories of local modules of type D (Q2224468)

Let \(\mathscr{C}(\mathfrak{g}, k)\) is the braided fusion category of finite-dimensional integrable representations of the quantized enveloping algebra \(U_q(\mathfrak{g})\). Here \(\mathfrak{g}\) is a complex finite-dimensional simple Lie algebra; \(k\) is a positive integer, called the level; \(q\) is a root of unity determined uniquely by \(\mathfrak{g}\) and \(k\). Let \(R\) be the algebra of complex functions on a finte subgroup \(H\) of the center of simply-connected compact Lie group corresponding to \(\mathfrak{g}\). Here \(\text{Rep}(H)\) is required to be a Tannakian fusion subcategory of \(\mathscr{C}(\mathfrak{g}, k)_pt\), which is the maximal fusion subcategory of \(\mathscr{C}(\mathfrak{g}, k)_pt\) where all simple objects are invertible. The paper calls such an algebra \(R\) a connected \(\acute{\text{e}}\)tale algebra of type D. \(R\) is also viewed as an object in \(\mathscr{C}(\mathfrak{g}, k)\). The paper studies the tensor category \(\mathscr{C}(\mathfrak{g}, k)_R^0\), consisting of local right \(R\)-modules, i.e., the structure map \(\rho: M\otimes R \to M\) satisfies \(\rho c_{R, M} c_{M, R} = \rho\), where \(c_{R, M}: R\otimes M \to M\otimes R\) and \(c_{M, R}: M \otimes R \to R \otimes M\) are the braiding operators. The results are organized by the Dynkin type of \(\mathfrak{g}\). \begin{itemize} \item[1.] If \(\mathfrak{g}\) is of type \(E_8, F_4\) and \(G_2\), then every \(R\) is trivial and there is nothing to say. \item[2.] when \(\mathfrak{g}\) is of type \(B_n, C_n, E_6\) or \(E_7\), Theorem 1 states that for every \(R\), \(\mathscr{C}(\mathfrak{g}, k)_R^0\) is simple (i.e., it contains no nontrivial proper fusion subcategories), except for the cases \(\mathscr{C}(\mathfrak{so}_{2n+1}, 2)\), \(\mathscr{C}(\mathfrak{so}_{2n+1}, 4)\) and \(\mathscr{C}(\mathfrak{sp}_{8}, 2)_R^0\) (here \(2n+1\) is not a prime). \item[3.] When \(\mathfrak{g}\) is of type \(A_n\) or \(D_n\), \(\mathscr{C}(\mathfrak{g}, k)_R^0\) is not simple. The paper discusses all the nondegenerate fusion subcategory of \(\mathscr{C}(\mathfrak{g}, k)_R^0\). Theorem 2 shows that such a subcategory is either pointed or the relative commutant of a pointed fusionn subcategory, except for the following cases: \begin{itemize} \item[a)] \(\mathscr{C}(\mathfrak{so}_{2n}, 2)_R^0\) where \(n\geq 5\) and \(R\) corresponds to an off-diagonal subgroup of \(Z_\mathfrak{g}\). \item[b)] \(\mathscr{C}(\mathfrak{so}_{4n}, 2)_R^0\) where \(n\geq 4\) and \(R\) is maximal (dimension 4). \end{itemize} \end{itemize} Two nondegenerate braided fusion categories \(\mathscr{C}_1\) and \(\mathscr{C}_2\) are Witt equivalent, if there exists two fusion categories \(\mathscr{A}_1\) and \(\mathscr{A}_2\), such that the categories \(\mathscr{C}_1\boxtimes\mathscr{Z}(\mathscr{A}_1)\) and \(\mathscr{C}_2\boxtimes\mathscr{Z}(\mathscr{A}_2)\) are braided equivalent. Here \(\mathscr{Z}(\mathscr{A})\) is the Drinfeld center of \(\mathscr{A}\). The set of Witt equivalence classes of nondegenerate braided fusion category forms a commutative group with the \(\boxtimes\) operation, called the Witt group. As an application of Theorem 1 and Theorem 2, the paper classifies relations in the Witt group generated by the equivalence classes of \(\mathscr{C}(\mathfrak{so}_5; k)\) and \(\mathscr{C}(\mathfrak{g}_2, k)\) for \(k \geq 1\).