Strange Duality of Verlinde spaces for G_2 and F_4

Abstract

We prove that the pull back of the canonical theta divisor for E_8-bundles at level one induces a strange duality between Verlinde spaces for G_2 and F_4 at level one on smooth curves of genus g. We also prove a parabolic generalization in terms of conformal blocks and write down identities between conformal blocks divisors in the Picard group of \bar-M_{g,n}.

Suggestions and comments are welcome

References in corpus (3)

• The rank-level duality for non-abelian theta functions
• The Chern character of the Verlinde bundle over the moduli space of stable curves
• The Chern classes of the Verlinde bundles