Affine Kac-Moody algebras graded by affine root systems (Q1849118): Difference between revisions

An affine Lie algebra \(\widehat{\mathfrak g}\) over \({\mathbb C}\) is said to be graded by the affine root system \(\Sigma\) if \(\widehat{\mathfrak g}\) contains an affine subalgebra \({\mathfrak a}\) (the `grading subalgebra'), whose root system relative to a Cartan subalgebra \(\mathfrak{h_a}\) is \(\Sigma\) and such that \(\widehat{\mathfrak g}=\oplus_{\lambda\in\Sigma\cup\{0\}}V_\lambda\); where \(V_\lambda=\{x\in \widehat{\mathfrak g}: [h,x]=\lambda(h)x\;\forall h\in\mathfrak{h_a}\}\). This extends the definition of Lie algebras graded by finite root systems given by \textit{S. Berman} and \textit{R. V. Moody} [Invent. Math. 108, 323-347 (1992; Zbl 0778.17018)]. This paper is devoted to the classification of the affine Lie algebras graded by affine root systems. The methods used are based on and extend those used in the classification by the same author [J. Algebra 223, 307-343 (2000; Zbl 0981.17023)] of the simple finite dimensional Lie algebras graded by finite root systems. Given a root basis \(\Pi\) of the root system of an affine Lie algebra \(\widehat{\mathfrak g}\), some specific subsets \(\Gamma\) of \(\Pi\) play a key role. For any such pair \((\Pi,\Gamma)\), called a \(C\)-admissible pair, a distinguished affine subalgebra \(\widehat{\mathfrak g}_\Gamma\) is defined naturally, so that \(\widehat{\mathfrak g}\) is graded over the root system of \(\widehat{\mathfrak g}_\Gamma\), with this latter subalgebra as the grading subalgebra. Now, if \(\widehat{\mathfrak g}\) is \(\Sigma\)-graded with grading subalgebra \({\mathfrak a}\), there exists a \(C\)-admissible pair \((\Pi,\Gamma)\) such that \({\mathfrak a}\) is contained in \(\widehat{\mathfrak g}_\Gamma\) and \(\widehat{\mathfrak g}_\Gamma\) is \(\Sigma\)-graded too. The grading is said to be `maximal' if \(\widehat{\mathfrak g}=\widehat{\mathfrak g}_\Gamma\). Moreover, the \(\Sigma\)-grading thus obtained in \(\widehat{\mathfrak g}_\Gamma\) is always maximal. This splits the problem of classifying the gradings into two pieces: to find all \(C\)-admissible pairs of a given root basis \(\Pi\), and to find the maximal gradings. Both problems are fully solved and the solution to each of them is presented in a table containing all the possibilities in terms of extended Dynkin diagrams with some extra information.