ffc team mailing list archive

[Patrick Riesen <priesen@xxxxxxxxxxxxxxx>]

> > due to the powerlaw behavior if have dot-products which have the form as 
> > (here v=test,u=trial,u0=func)
> >
> > dot(|grad(u0)|*|grad(u0)| * grad(u), grad(v))
> >
> > where |grad(u0)| is dot(grad(u0),grad(u0))
> >
> > so using
> >
> > M=dot(grad(u0),grad(u0))*dot(grad(u0),grad(u0))*dx
> >
> > in another form-file should may work?
>
> Yes, it should work, and the compilation and runtime performance of your code
> will benefit from it. The downside of course is that the user will have to
> handle the additional forms etc.

ok, i will pursue this approach. and do the the newton solving and 
computing of the terms in the increments accordingly. let's see how far 
i get before stumbling.

thanks for your help, kristian,

regards,
patrick