Primitive Reconstruction, Primitive Averaging

This is the first implementation. The update of each step ( update.f) reconstructs the primitive variables, for each one of course, and then the flux is calculated using these variables. For this we call getflux.f. This subroutine average on the primitive variables and then performs the spectral decomposition using these averaged values $(PA,\rho A,vA)$. Now we calculate conservative variables in terms of the left and right primitive variables from the reconstruction. Next we calculate the flux components and perform the Roe approximation. The control goes back to update.f and updates for fluxes and sources are performed.

We observe that for model A we have some roundness in the corners of the three variables, as well as for the B model. For C we have some smoothness and some transitorious events, mainly in the density evolution. They may not have physical meaning and could be numerical effects, because we have no floor [10].