8from .kernels
import create_compute_kernel_for_FD4
24 name, patch_size, rk_order, unknowns, auxiliary_variables, min_meshcell_h, max_meshcell_h, time_step_relaxation,
25 reconstruction_with_rk=False,
26 flux=PDETerms.User_Defined_Implementation,
27 ncp=PDETerms.None_Implementation,
28 point_source=PDETerms.None_Implementation,
29 boundary_conditions=PDETerms.User_Defined_Implementation,
30 refinement_criterion=PDETerms.Empty_Implementation,
31 initial_conditions=PDETerms.User_Defined_Implementation,
32 source_term=PDETerms.None_Implementation,
33 eigenvalues=PDETerms.User_Defined_Implementation,
34 pde_terms_without_state: bool =
False,
35 plot_grid_properties=
False, KOSigma=8.0
42 time_step_relaxation: Float
43 Calibration factor of CFL condition. The Runge-Kutta order is multiplied
44 with this value following the formula for Cockburn-Shu damping. However,
45 also the actual polynomial order has to enter the chosen time step size
46 through an additional @f$ p^{-2} @f$ scaling. We expect the user to
47 incorporate such an additional scaling within time_step_relaxation.
52 if reconstruction_with_rk:
57 super(GlobalAdaptiveTimeStepWithEnclaveTasking,self).
__init__(name,
66 kernel_namespace=
"fd4",
67 pde_terms_without_state=pde_terms_without_state,
77 eigenvalues=eigenvalues,
78 boundary_conditions=boundary_conditions,
79 refinement_criterion=refinement_criterion,
80 initial_conditions=initial_conditions,
81 source_term=source_term )
85 flux=None, ncp=None, source_term=None, eigenvalues=None,
86 boundary_conditions=None,refinement_criterion=None,initial_conditions=None,
87 memory_location = None,
88 additional_action_set_includes = "",
89 additional_user_includes = "",
93 If you pass in User_Defined, then the generator will create C++ stubs
94 that you have to befill manually. If you pass in None_Implementation, it
95 will create nop, i.e. no implementation or defaults. Any other string
96 is copied 1:1 into the implementation. If you pass in None, then the
97 set value so far won't be overwritten.
99 Please note that not all options are supported by all solvers.
101 This routine should be the very last invoked by the constructor.
104 flux, ncp, source_term, eigenvalues,
105 boundary_conditions, refinement_criterion, initial_conditions, memory_location, additional_action_set_includes, additional_user_includes)
107 if not KOSigma==
None: self.
_KO_Sigma = KOSigma
113 compute_max_eigenvalue_of_next_time_step =
True,
114 solver_variant = SolverVariant.WithVirtualFunctions,
115 kernel_variant = KernelVariant.PatchWiseAoSHeap,
122 compute_max_eigenvalue_of_next_time_step =
True,
123 solver_variant = SolverVariant.Stateless,
124 kernel_variant = KernelVariant.PatchWiseAoSHeap,
133 compute_max_eigenvalue_of_next_time_step =
True,
134 solver_variant = SolverVariant.AcceleratorWithExplicitCopy,
135 kernel_variant = KernelVariant.PatchWiseAoSHeap,
161 return super(GlobalAdaptiveTimeStepWithEnclaveTasking, self).user_action_set_includes +
"""
162#include "exahype2/fd/fd4/FD4.h"
168 Invoce superclass and then add one more field:
170 - RECONSTRUCTION_STAGES
Code snippet generator for fixed time stepping in the Runge-Kutta schemes.
_abstract_solver_user_definitions
_finish_time_step_implementation
_start_time_step_implementation
_compute_new_time_step_size
_solver_user_declarations
_abstract_solver_user_declarations
create_action_sets(self)
Create required action sets.
_constructor_implementation
Enclave variant of the solver where we still run through mesh once per Runge-Kutta sweep.
_fused_compute_kernel_call_gpu
_fused_compute_kernel_call_cpu
create_action_sets(self)
Call superclass routine and then reconfigure the update cell call.
set_implementation(self, flux, ncp, source_term, eigenvalues, boundary_conditions, refinement_criterion, initial_conditions, memory_location, additional_action_set_includes, additional_user_includes)
If you pass in User_Defined, then the generator will create C++ stubs that you have to befill manuall...
_eigenvalues_implementation
create_action_sets(self)
Call superclass routine and then reconfigure the update cell call.
_source_term_implementation
_solver_user_declarations
set_implementation(self, flux=None, ncp=None, source_term=None, eigenvalues=None, boundary_conditions=None, refinement_criterion=None, initial_conditions=None, memory_location=None, additional_action_set_includes="", additional_user_includes="", KOSigma=None)
If you pass in User_Defined, then the generator will create C++ stubs that you have to befill manuall...
_abstract_solver_user_definitions
_constructor_implementation
_fused_compute_kernel_call_gpu
__init__(self, name, patch_size, rk_order, unknowns, auxiliary_variables, min_meshcell_h, max_meshcell_h, time_step_relaxation, reconstruction_with_rk=False, flux=PDETerms.User_Defined_Implementation, ncp=PDETerms.None_Implementation, point_source=PDETerms.None_Implementation, boundary_conditions=PDETerms.User_Defined_Implementation, refinement_criterion=PDETerms.Empty_Implementation, initial_conditions=PDETerms.User_Defined_Implementation, source_term=PDETerms.None_Implementation, eigenvalues=PDETerms.User_Defined_Implementation, bool pde_terms_without_state=False, plot_grid_properties=False, KOSigma=8.0)
Construct solver.
user_action_set_includes(self)
Add further includes to this property, if your action sets require some additional routines from othe...
_compute_new_time_step_size
_start_time_step_implementation
_abstract_solver_user_declarations
add_entries_to_text_replacement_dictionary(self, d)
Invoce superclass and then add one more field:
_source_term_implementation
_fused_compute_kernel_call_cpu
_finish_time_step_implementation