[Developers] Evolving grid arrays in time

Ian Hawke I.Hawke at soton.ac.uk
Mon Sep 4 04:07:18 CDT 2006


On Tue, 2006-08-29 at 19:19 -0500, Erik Schnetter wrote:
> It is currently not possible to evolve grid arrays in time with MoL  
> when using mesh refinement.  At the moment, MoL would evolve grid  
> arrays multiple times per time step -- once for each refinement  
> level.  I am toying with different ideas to make Carpet, MoL, and  
> grid arrays work together.

> MoL would need additional code to ensure that grid arrays are evolved  
> only in global mode.
> 
> When grid functions are evolved in time, one cannot access the values  
> of grid arrays which are also evolved in time.  This is so because  
> their time stepping is non commensurable.
> 
> Comments?

This final point was the thing that made me think that it wasn't worth
doing, as there are few tasks that I could think of fitting this scheme
that couldn't be done with much less effort by post-processing. I would
only see the value if the evolved grid array was going to be used in the
evolution, and hence their values will need to be accessed.

It might be possible to allow this with a careful use of timelevels, but
that would require quite a bit of communication between Carpet and MoL I
would think.

Ian





More information about the Developers mailing list