|
Uni20
Asynchronous tensor and runtime library.
|
Stopping policy; tolerance is expressed in the residual real scalar. More...
#include <refinement_info.hpp>
Public Attributes | |
| std::optional< R > | backward_error_tolerance = {} |
| Componentwise backward-error target; absent means working-precision epsilon. | |
| std::size_t | max_iterations = 10 |
| Maximum correction steps per RHS, excluding the initial solve. | |
Stopping policy; tolerance is expressed in the residual real scalar.
| std::optional<R> uni20::linalg::RefinementOptions< R >::backward_error_tolerance = {} |
Componentwise backward-error target; absent means working-precision epsilon.
| std::size_t uni20::linalg::RefinementOptions< R >::max_iterations = 10 |
Maximum correction steps per RHS, excluding the initial solve.