Background
A 2-D slice from a 3-D simulation evolving the standard WaveToy example found in the Cactus distribution. No changes were made to the existing Cactus thorns to achieve mesh refinement. Only a few parameters in the parameter file were changed to activate mesh refinement:
Taka is a CACTUS driver, like PUGH, which supports parallel adaptive mesh refinement. It uses the SAMRAI toolkit and scales up to 100's of processors. Using Taka requires little or no modification of an existing CACTUS application. Only the thornlist in the parameter file must be modified ( if you are willing to accept the default interpolation and refinement criteria -- otherwise, you will need to supply some thorns to do this. Examples are provided ).
Example:
Change Active Thorns: from pugh to Taka
Add parameter: driver::max_no_of_levels = 2 (will support arbitrary # of levels )
Add parameter: driver::regrid_every = 10
By default, the driver refines on amplitude.
Taka currently supplies only refinement in space, not time. The timestep is consequently limited according to the resolution of the finest grid.
Parallel IO is already supplied as well as the standard CACTUS reduction operations. IO can be visualized without any post-processing using DataVault. ChomboVis and native OpenDX formats are also projected.