Multiplets of representations, twisted Dirac operators and Vogan's conjecture in affine setting (Q2480299)

In [Proc. Natl. Acad. Sci. USA 95, No. 15, 8441--8442 (1998; Zbl 0918.17002)], \textit{B. Gross, B. Kostant, P. Ramond} and \textit{S. Sternberg} studied certain multiplets of finite-dimensional representations of compact Lie groups and proved a number of their properties. In [Duke Math. J. 100, No. 3, 447--501 (1999; Zbl 0952.17005)] \textit{B. Kostant} defined a cubic Dirac operator \(D\) corresponding to a quadratic subalgebra of a (finite-dimensional) Lie algebra and showed how the multiplets arise as the kernel of \(D\). In a series of lectures [Dirac operator and unitary representations, 3 lectures at MIT Lie group seminar, Fall of 1997], \textit{D. Vogan} introduced the notion of Dirac cohomology of Harish-Chandra modules and formulated a conjecture about infinitesimal character of irreducible modules having nonzero Dirac cohomology. This conjecture was proved by \textit{J.-S. Huang} and \textit{P. Pandžić} in [J. Am. Math. Soc. 15, No. 1, 185--202 (2002; Zbl 0980.22013)]. The conjecture and its proof were then extended to the setting of the cubic Dirac operator by \textit{B. Kostant} in [Prog. Math. 210, 69--93 (2003; Zbl 1165.17301)]. The last paper contains also other results, notably nonvanishing of Dirac cohomology for certain highest weight modules. The aim of the present paper is to find analogs of these notions and results in the setting of affine Lie algebras and twisted affine cubic Dirac operators. There are also some applications, like strengthening the results about multiplets in the classical setting, obtaining the ``very strange formula'' and generalizing some results of \textit{G. D. Landweber} [Duke Math. J. 110, No. 1, 121--160 (2001; Zbl 1018.17016)]. Since affine Lie algebras and their representations are much more complicated objects then finite-dimensional Lie algebras, their finite-dimensional representations, and Harish-Chandra modules, it is not surprising that the analogs in the affine setting are not straightforward to formulate, and are more complicated to state and prove. For example, in order to work effectively with the center, one has to use completion, which brings in certain difficulties. Also, to talk about dimensions of representations in a multiplet, one has to employ the notion of asymptotic dimension, as the ordinary dimensions are infinite. The calculations use the machinery of vertex algebras and Lie conformal algebras. The paper starts with a quick review of basics about affine, fermionic and super affine vertex algebras and their twisted representations. After that, the (twisted affine) Dirac operator is introduced, then the results about multiplets of representations are discussed. After applying these to the very strange formula, the rest of the paper is devoted to formulating and proving the analog of Vogan's conjecture.