An application of Lie superalgebras to affine Lie algebras (Q753940)

Let \({\mathfrak g}\) be a finite dimensional complex simple Lie algebra and let \(\hat{\mathfrak g}\) be the associated affine Kac-Moody Lie algebra. Chari and Chari-Pressley classified the irreducible unitary representations of \(\hat{\mathfrak g}\) with finite dimensional weight spaces. Their result is that these are either integrable highest or lowest weight modules or certain ``loop modules with parameters''. After their work, it was natural to ask what are the irreducible unitary modules with infinite dimensional weight spaces. In an earlier work, they gave a class of such modules by showing that the tensor product of an integrable highest weight \(\hat{\mathfrak g}\)-module \(X(\lambda)\) with a loop module \(L(V(u))\) (where \(V(u)\) is a finite dimensional irreducible representation of \({\mathfrak g}\) with highest weight u) is irreducible (and clearly unitary) if \(\lambda\) is ``small'' compared to u. The main result of the paper under review (stated below) complements this result: The tensor product \(X(\lambda)\otimes L(V(u))\) is reducible (i.e. is not irreducible) if \(\lambda (c-o^{\vee})\geq D(V(u))\), where c is the `canonical' central element of \(\hat{\mathfrak g}\), \(o^{\vee}\) is the highest co-root of \({\mathfrak g}\) and \(D(V(u)):=\frac{2\Omega (V(u))\cdot \dim (V(u))Cox({\mathfrak g})}{\dim ({\mathfrak g})}\) is the Dynkin index of \(V(u)\); where \(Cox({\mathfrak g})\) is the dual Coxeter number of \({\mathfrak g}\) and \(\Omega(V(u))\) is the scalar by which the `normalized' Casimir operator acts on \(V(u)\). The proof makes use of a Lie superalgebra constructed from \(\hat{\mathfrak g}\) and \(V(u)\), which generalizes a construction due to Kac-Todorov.