dolfin team mailing list archive

[Johan Hake <johan.hake@xxxxxxxxx>]

[snip]

> Just as a side note, we first consider changing the meaning of
> TrialFunction to be a special subclass of Expression 

Should that be Coefficient?

> which (a Function
> in DOLFIN) instead of an Argument. So one would always define a
> nonlinear problem and it would always be linearized. 

Sounds like FEtk ;)

> That would solve
> some of the issues regarding the need for redefining u when a linear
> problem is solved:
> 
> u = TrialFunction(V)  # now it's a trial function
> ...
> u = problem.solve()   # now it's something different
> 
> But we decided this would require too many changes.

Yes, but I get the impression that you want to something like that but 
preserve backward compatability?

Johan