Tests with line scattering

We have run a number of test calculation similar to the LTE case but with line scattering included. In Fig. 3 we show the results for $\epsilon _l=10^{-4}$ and in Fig. 4 we show the results for $\epsilon _l=10^{-8}$ as examples. In both cases, the dynamical ranges of ${\bar J}$ are much larger than in the LTE case. The 3D calculations compare very well to the 1D calculations, in particular in the outer zones. In the inner parts the resolution of the 3D models is substantially lower than for the 1D models, therefore the differences are largest there. In Figs. 3 and 4 we show the results for a test model with a grey temperature structure. In these models, the 3D spatial grid was substantially larger ( $n_x=n_y=n_z=2*96+1$) in order to help resolve the inner regions where the temperature gradients are very large. As expected, the agreement is not as good in the inner regions as in the models with a constant temperature, however, the models agree very well in the outer regions. Overall, the agreement is very similar in quality compared to the case with constant temperatures. The differences in the mean intensity $\bar J$ between the 1D comparison case and the 3D case is several per-cent in the innermost layers where the grid of the 3D case is under-sampled, the differences are below 0.1% in the outer zones.