Re: [pooma-dev] Chevron Flow Scenario

[Scott Haney]

On Wednesday, April 11, 2001, at 04:25 PM, Jeffrey Oldham wrote:

> q = dot( dot( replicate<2>(K), \grad(P)), replicate<2>(normals));

Hi Jeffrey,

In 2D there are 4 values of \grad(P) and K (after replication) per cell. 
The inner dot makes sense.

I don't know fully what it means to replicate an all-face centered field 
like normals, but I'll assume it means to double the number of normals 
in each direction. Then, I don't understand the counting since there are 
8 normals per cell in 2D and 24 in 3D, compared to 4 and 8 cell-centers 
(on the sub-mesh) per cell respectively. This illustrates my primary 
concern with the proposal as currently written. It is centering-free and 
don't think you can get things to work out in general unless you specify 
this information. I also don't think we need to replicate values if we 
are  precise about which one's we're talking about.

Scott