On 9/26/07, Nicolas Neuss <address@hidden> wrote:
"Jason Sidabras" <address@hidden> writes:
> I am still learning what I have to solve but I believe that is close. The
> current plan is to formulate the equation based off of the electric
> vector potential (A) and scalar magnetic potential (phi).
Shouldn't this be the other way round? Magnetic vector potential and
electric scalar potential?
> curl (sigma^-1 curl A) = -i omega mu (A - div phi)
Divergence of a scalar potential does not make sense.
ugh, how embarrassing.
curl (sigma^-1 curl A) = -i omega mu (A - grad phi)
I am currently basing my knowledge off a few papers and commercial FEM simulations. The equation above comes from :
Jiabing, W., Baoqian, T. "Calculation of 3d eddy current problems using a modified T-Ohmega Method", IEEE Trans on Magn, Vol. 24, No. 1 (Jan. 1988)
The domain is a non-conducting region with a conducting subdomain inside. The above equation is used for the conducting region.
I believe an electric vector potential is used because:
curl A = J
and J is the eddy current on the conducting subdomain.
Possible references: the book of Monk, two recent books of Demkowicz, and
maybe also an Acta Numerica article of Hiptmair.
Thanks, I will check them out.
The file triangle.lisp is not part of the Shewchuk library, but the Femlisp
interface to this library. The function TRIANGULATE-2D writes out a
boundary polygon, then calls 'triangle', then reads in the generated mesh.
I found it. Using (fl.mesh::triangulate-2d x) instead of (fl.mesh:triangulate-2d x)
with the :: it shows up in my slime-repl
Thanks for everything,
Jason
HTH, Nicolas