ZhiChao
The radius specified in the parameter file is the radius of the "interior" of that refinement level. First, Carpet rounds it up to the next grid point if necessary. Then, Carpet will add buffer points and ghost points to the refinement level. These points are not really evolved, they are interpolated from the next coarser grid. Ideally, these points would not be displayed, but I don't think the Carpet-VisIt interface is clever enough to do that.
If you are using an RK4 integrator and are using 3 ghost points, then this would add 12 points to each level. On the coarsest level, this would change the radius from 270 to 270+38.4 = 308.4. This is still less than the 352 you report. I don't know where the rest of the difference comes from.
-erik
On Thu, Apr 30, 2020 at 11:30 AM ZhiChao Zhao yanyuechuixue@gmail.com wrote:
Hi All,
I am using ET to simulate a BNS merge. I set the radius as(max 7 level):
CarpetRegrid2::radius_1[1] = 270.0 CarpetRegrid2::radius_1[2] = 162.0 CarpetRegrid2::radius_1[3] = 94.5 CarpetRegrid2::radius_1[4] = 40.5 CarpetRegrid2::radius_1[5] = 27.0 CarpetRegrid2::radius_1[6] = 13.5
However, when I check the radius of mesh in VisIt(the output h5 file), I found the radius of each level becomes: [ 512, 352, 200, 112, 50, 32, 16] , which is different from what I have written in CarpetRegrid2:: radius_1[index].
Could anyone tell me how the radius transfer from 270.0 to 352? Or give me any document of Carpet?
My another parameters are:
CoordBase::dx = 6.4 CoordBase::xmax = 512.00 Carpet::ghost_size = 3 Carpet::prolongation_order_space = 5
Thanks a lot! Yours, Zhichao Zhao.
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users