Date: Aug 5, 2009 6:21 PM
Author: Bruno Luong
Subject: Re: measured boundary conditions with pde toolbox

"Doug " <dhk@umd.edu> wrote in message <h5cmv4$lbc$1@fred.mathworks.com>...
> Has anybody tried using the pde toolbox with boundary conditions taken from experimental measurements (ie, not a formula)?
>
> I see that assempde will take as its input either a boundary condition matrix or a boundary M-file, but both seem to give boundary conditions by specifying a simple formula (like x.^2+y.^2, from the example in the help files). I want to solve Poisson's equation using actual experimental data for the Neumann boundary conditions. It would seem reasonable to provide a matrix similar in form to the edge matrix "e", but containing local values of the Neumann boundary condition. Can it be done?
>
> Thanks in advance from a lowly experimentalist to all you pde gurus!


Unless if you refer to Dirac Neumann bc, the local values Neumann's bc does not make sense. It must be something you can *apply* on another function defined on the boundary. Specifically, mathematician like to refer it as continuous linear application on the fractional Sobolev H^{1/2} space; or even more barbaric: the "H^{-1/2} space". The later does not have defined local value, unfortunately.

Bruno