search

fastalgorithms / fmm3dbie

GNU General Public License v2.0
25 stars 10 forks source link

helm.kern 'eval' interface is not the CFIE #19

Closed ahbarnett closed 3 months ago

ahbarnett commented 3 months ago

I don't understand the point of the eval option to helm3d.kern It seems to stack [aS, D] output matrices where a is the scale factor in varargin{1}. But why not return aS+D as in CFIE?

mrachh commented 3 months ago

The 'eval' routine is meant for things like transmission problems, or even Neumann problems, where you have two different densities for the single and double layer potentials, as opposed to 'comb' or 'c' which is meant to be a combined field representation where S and D have the same density. It could use a better name, but since it arises both in transmission and Neumann problems, we didn't want to commit to either of those options.

ahbarnett commented 3 months ago

ok, got it.