We try to understand why the same "diophantine condition" appears both in the study of the conjugation of the diffeomorphisms of the circle, and in that of the Hochschild cohomology of some crossed product algebras.