[CIG-LONG] Problem about the Mesh in Gale

[Walter Landry]

feng lin <rimbong at gmail.com> wrote:
> Hey everyone,
> 
> I'm now running some model to simulate some extension precess with Gale2.0.
> And I met some problem with the number of elements in the model.
> 
> I set up a 2D problem and in the input file:
> 
> {
> ...
>     "nx": "16",
>     "ny": "4",
> ...
> }
> 
> When I plot the v-mesh and pressure-mesh with VisIt2.5. I got both 32
> elements at x direction and 8 elements at y direction.
> 
> I can understand that for the v-mesh the number elements in the x direction
> should be 2*nx (32), y direction is 2*ny (8). The v -mesh is quadratic
> elements (Q2) according to the manual.
> But I couldn't understand why for the pressure-mesh the number of elements
> in x and y direction are 2*nx (32), and 2*ny (8). As the pressure mesh
> should be discontiuous linear elements(P-1) according to the manual, I
> think the number of elements in x, y direction should be nx(16), xy(8)

The pressure elements are not isoparametric.  So the interpolation
function (piecewise linear) is different from the mapping function
from the canonical element (Q2).  So pressure elements can be plotted
on the same mesh as the Q2 velocity elements, but they have fewer
degrees of freedom.

Cheers,
Walter Landry