Conflict between symmetries: a classical example

We have found a (classical) competition between duality and gauge symmetries when trying to obtain an explicit dual to the non-nonabelian version of the self-dual model proposed by Townsend, Pilch and van Nieuwenhuizen\cite{TPvN} (NASD) either through gauge embedding procedures or gauge invariant master approach. We found that the theory dual to NASD {\it is not} the Yang-Mills-Chern-Simons model. We then proved that a model gauge invariant and dual equivalent to NASD cannot be achieved because of this conflict. Other nonabelian self-dual formulation proposed in \cite{KLRvN} was studied and its dual obtained. We discuss the consequences of the dual equivalence found here in the context of 3D non-abelian bosonization