We present a general criteria to prove that a probability measure satisfies a logarithmic Sobolev inequality, knowing that some of its marginals and associated conditional laws satisfy a logarithmic Sobolev inequality. This is a generalization of a result by N. Grunewald et al. [N. Grunewald, F. Otto, C. Villani, M.G. Westdickenberg, A two-scale approach to logarithmic Sobolev inequalities and the hydrodynamic limit, Ann. Inst. H. Poincaré Probab. Statist., in press].The motivation behind this work is molecular dynamics (in the canonical statistical ensemble), and more precisely, (i) the analysis of numerical methods for the computation of free energy differences [8] (see Remark 1 below) and (ii) the derivation of effective dynamics on coarse-grained variables [7]. In both cases, it appears that estimates based on entropies for measures related to the Boltzmann-Gibbs measure is a useful tool. One important question is the following: what is the link between the logarithmic Sobolev inequality (LSI) constant of the Boltzmann-Gibbs measure for the original variables (microscopic level) and the LSI constant of the Boltzmann-