The trac suggests that there exists a "brutish" way of coaking fipy to
simultaneously solve multiple equations.

Are there any examples tucked away somewhere about how one might
proceed to do this?

I am trying to solve a set of non-linear PDE's and am trying to implement
the Newton step which requires solving a system

nabla^2 dF + dot(M(x), dF)

where dF = [df1, df2, df3,...] is a vector of steps and M(x) is the Jacobian
matrix of the non-linear source (depends on position).

Allowing for CellVariables to be vectors and Sources to be matrices
would also work (but does not seem possible).

The equations end up being quite strongly coupled (due to the
matrix M(x)), so iterating over each equation does not work very well.

Thanks for any suggestions,
Michael.


Reply via email to