LinearSolvers

There are five examples in the ExampleCodes/LinearSolvers directory.

ABecLaplacian_C demonstrates how to solve with cell-centered data in a C++ framework. This example shows how to use either hypre or PETSc as a bottom-solver (or to solve the equation at the finest level if you set the “max coarsening level” to 0.

Step-by-step instructions for configuring AMReX to use HYPRE for this example are available here.

ABecLaplacian_F demonstrates how to solve with cell-centered data using the Fortran interfaces.

GMRES/Poisson demonstrates how to solve the Poisson equation using GMRES with Jacobi preconditioning.

NodalPoisson demonstrates how to set up and solve a variable coefficient Poisson equation with the rhs and solution data on nodes.

MultiComponent demonstrates how to solve using a multi-component operator with nodal data. The operator is of the form $D(\varphi {)}_i={\alpha }_{ij}{\mathrm{\nabla }}^2{\varphi }_j$, where $\varphi$ is the multi-component solution variable and ${\alpha }_{ij}$ is a matrix of coefficients. The operator (MCNodalLinOp) is implemented using the “reflux-free” method.

NodeTensorLap solves the equation $\mathrm{\nabla }\cdot (\sigma \mathrm{\nabla }\varphi )=\text{RHS}$, where $\varphi$ and $\text{RHS}$ are nodal MultiFabs and $\sigma$ is a tensor constant. It is assumed that the tensor is symmetric with rank equal to AMREX_SPACEDIM.