dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

Martin Sandve Alnæs wrote:
> On Fri, Feb 27, 2009 at 4:08 PM, Garth N. Wells <gnw20@xxxxxxxxx> wrote:
> > Martin Sandve Alnæs wrote:
> > > On Fri, Feb 27, 2009 at 4:03 PM, Garth N. Wells <gnw20@xxxxxxxxx> wrote:
> > > > Evan Lezar wrote:
> > > > > Hi
> > > > >
> > > > > I am working on some electromagnetic problems that require I model the
> > > > > electric field as a complex phasor.  When compiling user-defined
> > > > > functions I get an error message.
> > > > >
> > > > > For example, consider the following simple function:
> > > > >
> > > > > f = Function(V, '1.0j')
> > > > >
> > > > >
> > > > > when calling the JIT compiler, the log shows the following error (in the
> > > > > last couple of lines):
> > > > >
> > > > > dolfin_compile_function_934f16b21b1210d9244eba2ba7985a9a_wrap.cxx: In
> > > > > member function ‘virtual void
> > > > > dolfin::function_91ee55df622466fafbf3ba887ffb745e::eval(double*, const
> > > > > double*) const’:
> > > > > dolfin_compile_function_934f16b21b1210d9244eba2ba7985a9a_wrap.cxx:3221:
> > > > > error: cannot convert ‘double __complex__’ to ‘double’ in assignment
> > > > > error: command 'gcc' failed with exit status 1
> > > > >
> > > > > Does dolfin support complex valued functions?
> > > > No.
> > > >
> > > > If so, how do I go about
> > > > > getting them working?  And if not, how do I go about adding the
> > > > > functionality.
> > > > It would be a big task to make it seamless. The starting point would be
> > > > having FFC (or soon UFL) support complex functions. I thought about this
> > > > a few years ago but never got around to it.
> > Thinking about, it shouldn't be too hard to patch something together using
> > mixed elements.
>
> Won't be as user-friendly as it could be though, but perhaps you
> could define e.g. complex multiplication as python functions on top.

Agree, it wouldn't be user friendly. Making it user friendly would 
require a lot of work on UFL/FFC/DOLFIN.

Garth


> > > > UFL question: Does or will UFL support complex functions?
> > > No. And I don't think making such an addition before it has stabilised
> > > is an option, we can't change everything everywhere simultaneously.
> > OK. Would be good for the long-term goals list.
>
> Sure. I'll add it.
>
> Martin