Aux-component-aware variants of the native Schur kernels (coupling/schur/core/schur_condensation.hpp.
More...
#include <program_context.hpp>
Aux-component-aware variants of the native Schur kernels (coupling/schur/core/schur_condensation.hpp.
- condensed_schur_source_stepper.hpp). The native kernels read B_z from a DEDICATED B_z MultiFab at component 0; a compiled Program reads B_z straight from the System aux channel at an arbitrary component
c_bz, so these thin wrappers carry c_bz and otherwise REPRODUCE the native formulas verbatim (same LorentzEliminator B^{-1}, same coefficients, same centered gradient) – the native CondensedSchur path is untouched. Named functors (device-clean, nvcc cross-TU rule, like the native ones). epic ADC-399 / ADC-421. A_op = I + c*rho*B^{-1} per cell (eps_x/eps_y diag, a_xy/a_yx cross). Mirrors detail::SchurOperatorCoeffKernel but reads B_z from the aux at c_bz.
◆ operator()()
| POPS_HD void pops::runtime::program::detail::SchurOperatorCoeffKernelC::operator() |
( |
int |
i, |
|
|
int |
j |
|
) |
| const |
|
inline |
◆ aux
| ConstArray4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::aux |
◆ axy
| Array4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::axy |
◆ ayx
| Array4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::ayx |
output: cross terms a_xy, a_yx
| Real pops::runtime::program::detail::SchurOperatorCoeffKernelC::c |
◆ c_bz
| int pops::runtime::program::detail::SchurOperatorCoeffKernelC::c_bz |
◆ c_rho
| int pops::runtime::program::detail::SchurOperatorCoeffKernelC::c_rho |
◆ ex
| Array4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::ex |
◆ ey
| Array4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::ey |
output: eps_x, eps_y (diagonal of A)
| ConstArray4 pops::runtime::program::detail::SchurOperatorCoeffKernelC::s |
fluid state (rho at c_rho)
◆ th_dt
| Real pops::runtime::program::detail::SchurOperatorCoeffKernelC::th_dt |
The documentation for this struct was generated from the following file: