Closed jackhamel16 closed 3 years ago
There was a slight hold up in the final expression for computing the scattering matrix. It relies on trans(Vs) * inv(Z) = trans(Js) which is only true when Z is symmetric. Z is symmetric only when doing full galerkin (i.e. same bases and quadrature rules for test and source integrals). Even though Z isn't perfectly symmetric under full Galerkin (maybe just due to numerical error?) we can assume the relation above is true because the derivative of the S matrix is about just as accurate under that assumption as when not. So in the end, we can assume Z is symmetric under full galerkin testing.
All is added but I am going to quickly code in the direct and indirect ways of computing the WS matrix, Q, as an additional check.
Follow derivations in write up.
Test by approximating derivative with finite difference