CFD Online Discussion Forums - leastSquares gradient on tet meshes

I've been trying out a few gradient schemes recently on tet-meshes, of which I find the leastSquares scheme to be most accurate. However, I find that the results are quite awkward at boundaries. My test case is quite simple (tet-mesh in a box - see attached), and the driver code is given:

Code:

#include "fvCFD.H"



int main(int argc, char** argv)

{

#   include "setRootCase.H"

#   include "createTime.H"

#   include "createMesh.H"



    volScalarField alpha

    (

        IOobject

        (

            "alpha",

            runTime.timeName(),

            mesh,

            IOobject::NO_READ,

            IOobject::NO_WRITE

        ),

        (mesh.C() & vector(2,3,1))

    );



    // Set values for the field

    volVectorField gradAlpha = fvc::grad(alpha);



    Info << "Writing out the gradient" << endl;



    gradAlpha.write();



    return 0;

}

I would expect grad(alpha) to be a constant vector(2,3,1), which is mostly the case at interior cells, but not so at cells the boundary. Am I doing something wrong?

Attachments:
http://www.ecs.umass.edu/~smenon/tarballs/testGrad.gz
http://www.ecs.umass.edu/~smenon/tarballs/boxTestCase.gz

Thanks in advance.