Recent neutron scattering results [F. Demmel et al., Phys. Rev. B 73, 104207 (2006)] on the temperature dependence of de Gennes narrowing in liquid rubidium have stimulated a molecular dynamics (MD) study in the same temperature and density range. At the $k$ value of the first peak of $S\left(k\right)$, the MD results agree very well with experimental data of $\tilde{S}(k_{\max },\omega =0)$, $F(k_{\max },t=0)$, and longitudinal viscosity ${\eta }_L(k_{\max },\omega =0)$. Other transport properties, such as self-diffusion and shear viscosity, are also accurately reproduced. At $k=0$, on the other hand, the MD results significantly underestimate the experimental values of bulk viscosity and thermal conductivity. For the latter, this is a well known deficiency of models which do not explicitly take into account the electronic contribution to thermal exchanges. However, the large difference between MD and macroscopic experimental data for bulk viscosity casts some doubts on its indirect calculation from sound absorption data. This contradictory result, which presumably extends to all alkali metals, is discussed in the light of various theoretical models.

• Received 22 May 2007

DOI:https://doi.org/10.1103/PhysRevB.76.094204