Thank you again<br>I look forward to laying my hands on this project. But since I have to finish my thesis in the next two weeks or so you will probably hear not much more of me till then.;)<br>Besides I will have to learn a bit C++ before I start. I am already familiar with object oriented languages (ruby and java) but not with all that header file stuff and certainly not with templating. Maybe you could give me a hint about a good online-tutorial or book.<br>
But only if you have time.<br><br>Greetings <br>Markus<br><br><br><div class="gmail_quote">2008/6/28 Christian Engwer <<a href="mailto:christi@uni-hd.de">christi@uni-hd.de</a>>:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Dear Markus,<br>
<div class="Ih2E3d"><br>
> Thank you very much for the quick answer.<br>
> The altix has an intel 10.1 c++ compiler istalled and could be an option.<br>
> <a href="http://www.lrz-muenchen.de/services/software/programmierung/intel_compilers/" target="_blank">http://www.lrz-muenchen.de/services/software/programmierung/intel_compilers/</a><br>
> The proposed project migration would not be a mere porting but a complete<br>
> reimplementation of the solver which is a mixture of fortran 77 and 90 with<br>
> MPI up to now.<br>
<br>
</div>That should be fine.<br>
<div class="Ih2E3d"><br>
> So there is no code that could get in the way yet, but its a lot of work to<br>
> be done and we are anxious to get the answers to some questions before we<br>
> start. Perhaps you can answer some of them out of hand right now.<br>
> You mentioned some parallel grid managers. Since up to now I have written<br>
> the code needed for discretization myself, I do not know exactly what a tool<br>
> like a grid manager really does and, even more important, what it does not.<br>
> So I just describe our problem and what we hope to gain using dune.<br>
> We certainly need an inf-sup stable finite element pair( Taylor Hood for<br>
> instance) for a Stokes system with very strongly varying and even<br>
> discontinuous viscosity.<br>
> We also have to discretize a convection diffusion equation so wee need the<br>
> ability to implement a streamline diffusion discretization, possibly with<br>
> the same elements we used for the Stokes system.<br>
> We want to use a parallel algebraic multigrid solver for both of the<br>
> discretized problems (because of the discontinuities).<br>
> We want to be able to implement adaptive mesh refinement and do not want to<br>
> bother about load balancing issues arising from this.<br>
><br>
> So<br>
> 1.) We would be thankful if you could give us your opinion if you deem our<br>
> plan possible!<br>
<br>
</div>I think it is feasible. The stable Dune modules currently cover the<br>
grid interface and istl (the iterative solver template<br>
library). dune-istl also includes an algebraic multigrid solver.<br>
<div class="Ih2E3d"><br>
> and<br>
> 2.) perhaps tell me what grid managers or solvers are available and how I<br>
> can find out, what I need to know.<br>
> (A link to a documentation of a parallel grid manager that works with dune<br>
> perhaps)<br>
<br>
</div>OK. From the features page<br>
<a href="http://www.dune-project.org/doc/devel/features.html" target="_blank">http://www.dune-project.org/doc/devel/features.html</a><br>
you find links to the documentation of the different grid<br>
implementations. All parallel adaptive grid managers are third party<br>
software, but accessible thorugh the dune grid interface. The grid<br>
mangers takes care of the details of grid refinement and load<br>
balancing. It offers you a clear interface of how to access the<br>
entities in the grid. We two papers linked from our website, the first<br>
one gives clear definition of what a grid is (very mathematical) and<br>
the second one describes how this definition is transformed into a c++<br>
interface and shows some applications.<br>
<br>
Cheers Christian<br>
</blockquote></div><br>