ghost node positions

[AlexGlock]

Hello there,

I am experimenting with a self written 2D finite difference solver as a drop in replacment to the picmi.ElectrostaticSolver. I based my code on the capacitive discharge example.

My question is regarding the physical position of the ghost nodes/ guard nodes which are included in the amrex electric potential vector. Are they located within the limits of the cartesian grids or added as an outside layer with offset of dx or dz?

Hope it is kind of understandable what I mean.
Thanks in advance and greetings,

Alex

[ax3l]

Hi @AlexGlock,

That is a great question and probably under-documented. For the most outer blocks of the simulation the extra ghost cells are outside the "geometry" extends. So for instance, absorbing layers like PMLs are "outside" in the ghost/halo region (that AMReX calls nGrow or nGrowVect).

Best,
Axel