Mpdata's deformation tensor assumes incompressibility?

[pdziekan]

In the SGS model of libmpdata++ (calc_deform functions in stress_formulae.hpp), deformation tensor does not include velocity divergence on the diagonal. This is probably because it is assumed that the divergence is zero, what is not true in UWLCM. In UWLCM, anelastic approximation assumes that div(density*velocity)=0, not div(velocity)=0.

[pdziekan]

Update from Maciek Waruszewski: divergence is not included only in the standard formulation, in the compact formulation it is included. Testing shows that including the divergence has little effect (slightly increases SGS diffusion).