In this example, we will simulate the time evolution of a Gaussian hill in an infinite space with periodic boundaries. The initial form of the Gaussian pulse gives the form: The analytical solution with time is:
In our simulation, we define $nelem = 500$ which generates a 1000*1000 2D mesh. The mesh is recognized
as an infinite area for $\sigma_0=40$. The D2Q9 layout is used. Boundaries from all directions are
periodic. As the area is much larger than the $ \sigma_0 $, the effect from the boundaries
is neglected. Results of t=200 are used to calculate the diffusion factor.

In ./musubi.lua, define 'shape' in 'tracking' block as
shape = {
-- kind = 'all'
kind = 'canoND',
object = {
origin = {-nelem-1, 0, 0},
vec = { {2.0*(nelem+1), 0., 0.0} }
}
},
To generate a midline profile comparison between the analytical and computed solution, run profile.sh to create the plot.

In ./musubi.lua, define 'shape' in 'tracking' block as
shape = {
kind = 'all'
-- kind = 'canoND',
-- object = {
-- origin = {-nelem-1, 0, 0},
-- vec = { {2.0*(nelem+1), 0., 0.0} }
-- }
},
To generate the above plot, simply run test_stability.sh. Notice that when $ \tau = 1 $, the error of computed diffusion factor is exactly 0.

In test_order.sh, define $ \tau=0.501 $, and run test_order.sh. The chart shows the error of diffusion factor changing with the background velocity. The following charts can be obtained with different relaxation time $ \tau $.

The objectives of this example is to introduce how to: * Simulate time evolution of the advection-diffusion process of a 2D Gaussian Hill. * Compare the profile between first and second order bgk equilibria * Compute the diffusion parameter and compare it to the theoritical value for different background velocities