|
procedure, non_overridable | fd_scheme_ => scheme_construct |
| The default constructor will call init()
|
|
procedure, non_overridable | declare_stencil => scheme_declare_stencil |
| Declare a stencil base on its istarting and ending relative indices.
|
|
procedure, non_overridable | apply => scheme_apply |
| Apply a scheme to given discrete values.
|
|
procedure, non_overridable | init_fill => scheme_init_fill |
| Call the init() and fill() procedures.
|
|
procedure(fd_scheme_init), deferred, pass(self) | init |
|
procedure(fd_scheme_fill), deferred, pass(self) | fill |
| Init the scheme with its stencil. This has to be overriden by each scheme.
|
|
type(t_stencil_1d) | stencil |
| The weights are stored in the subtype stencilweight from index_start to index_end : fd_scheme%stencil%weight(fd_scheme%stencil%index_start:fd_scheme%stencil%index_end) .
|
|
integer | dir |
|
integer | f |
|
integer | the |
|
integer | direction |
|
integer | of |
|
integer | scheme |
|
integer | backward |
|
integer | or |
|
integer | forward |
|
- Warning
- the
fill
subroutine works differently: the first element of the array is the extrapolation step (where to extrapolate, starting from \(x(0)\)), the rest of the array is the discretization steps between the values \(\phi(x(i))\). This was done as the fill
is deferred (has to be defined) and we didn't want to build a new ext-scheme class.
-
The definition of the direction of the scheme is also different. Here, the backward term refers to where we extrapolate, hence the discretization points stands forward to the extrapolation destination.
- Note
reconstruction
is 0 (default) for interpolation and 1 for reconstruction
The documentation for this type was generated from the following file:
- src/lib/discretization/node_level_schemes/extrapolation/type_ext_scheme.f90