From Dynkin diagram symmetries to fixed point structures (Q1924094)

In this interesting paper the authors associate algebraic structures to automorphisms of Dynkin diagrams and study some of their interrelations. (For details of definitions and statements the original text should be consulted.) The class of Dynkin diagrams considered are those of symmetrizable Kac-Moody algebras. An automorphism of a Dynkin diagram is a permutation of its nodes which leaves the diagram invariant. They focus their attention on two main types of orbits, namely those where each of the nodes on an orbit is either connected by a single link to precisely one node on the same orbit or not linked to any other node on the same orbit. If all orbits of a given Dynkin diagram automorphism are of one of these two types, the automorphism satisfies the linking condition. Except for the order \(N\) automorphisms of the affine Lie algebras \(A^{(1)}_{N-1}\), all diagram automorphisms of simple and affine Lie algebras belong to this class. Any automorphism of a Dynkin diagram (not necessarily satisfying the linking condition) induces an outer automorphism of the associated Kac-Moody algebra \({\mathfrak g}\). For simple, affine and hyperbolic algebras the induced automorphism is unique. Further they define a new type of character-like quantities for these modules by inserting the generator of the automorphism into the trace that defines the ordinary character. They call the object constructed in this manner the twining character of the highest weight module. Trivially, the twining character vanishes whenever the highest weight is changed by the automorphism. As a consequence, the interest is in those highest weight modules whose highest weight is not changed by the automorphism; these special modules are called the fixed point modules of the automorphism and their highest weights are referred to as symmetric \({\mathfrak g}\)-weights. There is one interesting class of automorphisms for which only the highest weight state of a fixed point module contributes to the twining character. These are the order \(N\) automorphisms of the affine Lie algebras \(A^{(1)}_{N-1}\). In this particular case they prove that the Serre relations among the commutators of step operators conspire in such a way that all other states in the Verma module (and hence also in the irreducible module) cancel each others' contributions to the twining character. The main result of this paper concerns the fixed point modules of Dynkin diagram automorphisms which do satisfy the linking condition. The authors prove that these modules are in one-to-one correspondence with the highest weight modules of the orbit Lie algebra \(\breve {\mathfrak g}\), and that the twining characters of the fixed point modules (both for Verma modules and for their irreducible quotients) conicide with the ordinary characters of the highest weight modules of \(\breve {\mathfrak g}\). It is not claimed that the orbit Lie algebra \(\breve{\mathfrak g}\) is embedded in the original algebra \({\mathfrak g}\). They can show that the Weyl group of \(\breve {\mathfrak g}\) is isomorphic to a subgroup of the Weyl group of \({\mathfrak g}\). This observation plays a key rôle in the proof, as it enables to employ constructions that are analogous to those used by Kac in his proof of the Weyl-Kac character formula. In the final section the authors specialize to the case of untwisted affine Lie algebras and those automorphisms which correspond to the action of simple currents. The action of such automorphisms is described in detail, using the realization of affine Lie algebras as centrally extended loop algebras. They find that for this special class of automorphisms the characters of \(\breve {\mathfrak g}\), and hence also the twining characters, have nice modular transformation properties. Next it is shown that the modification of the irreducible characters of \(\breve {\mathfrak g}\), and hence of the irreducible twining characters of \({\mathfrak g}\), that is required in order to obtain these nice modular transformation properties, differs from the modification of the irreducible characters of \({\mathfrak g}\) only by an overall constant. Finally, comments are given on those cases where the orbit Lie algebra is one of the twisted affine Lie algebras \(B_n^{(2)}\) rather than an untwisted affine algebra. The main motivation for introducing and studying twining characters stems from a long-standing problem in conformal field theory, namely the ``resolution of fixed points''. Twining characters and orbit Lie algebras constitute important progress towards solving this problem. This has not been discussed further in the present paper, except for a brief explanation of the relation between the two issues. (Mainly according to the introduction).