#include <source/time_integrator.h>

Inheritance diagram for ryujin::TimeIntegrator< Description, dim, Number >:

Collaboration diagram for ryujin::TimeIntegrator< Description, dim, Number >:

Public Types
Typedefs and constexpr constants
using	HyperbolicSystem = typename Description::HyperbolicSystem

using	View = typename Description::template HyperbolicSystemView< dim, Number >

using	ParabolicSystem = typename Description::ParabolicSystem

using	StateVector = typename View::StateVector

Public Member Functions
Constructor and setup
	TimeIntegrator (const MPIEnsemble &mpi_ensemble, const OfflineData< dim, Number > &offline_data, const HyperbolicModule< Description, dim, Number > &hyperbolic_module, const ParabolicModule< Description, dim, Number > &parabolic_module, const std::string &subsection="/TimeIntegrator")

void	prepare ()

Functions for performing explicit time steps
void	prepare_state_vector (StateVector &state_vector, Number t) const

Number	step (StateVector &state_vector, Number t, Number t_final=std::numeric_limits< Number >::max())

const auto &	time_stepping_scheme () const

const auto &	efficiency () const

Number	step_ssprk_22 (StateVector &state_vector, Number t, Number tau_max)

Number	step_ssprk_33 (StateVector &state_vector, Number t, Number tau_max)

Number	step_erk_11 (StateVector &state_vector, Number t, Number tau_max)

Number	step_erk_22 (StateVector &state_vector, Number t, Number tau_max)

Number	step_erk_33 (StateVector &state_vector, Number t, Number tau_max)

Number	step_erk_43 (StateVector &state_vector, Number t, Number tau_max)

Number	step_erk_54 (StateVector &state_vector, Number t, Number tau_max)

Number	step_strang_ssprk_33_cn (StateVector &state_vector, Number t, Number tau_max)

Number	step_strang_erk_33_cn (StateVector &state_vector, Number t, Number tau_max)

Number	step_strang_erk_43_cn (StateVector &state_vector, Number t, Number tau_max)

Number	step_imex_11 (StateVector &state_vector, Number t, Number tau_max)

Number	step_imex_22 (StateVector &state_vector, Number t, Number tau_max)

Number	step_imex_33 (StateVector &state_vector, Number t, Number tau_max)

Detailed Description

template<typename Description, int dim, typename Number = double>
class ryujin::TimeIntegrator< Description, dim, Number >

The TimeIntegrator class implements IMEX timestepping strategies based on explicit and diagonally-implicit Runge Kutta schemes.

Definition at line 252 of file time_integrator.h.

Member Typedef Documentation

◆ HyperbolicSystem

template<typename Description , int dim, typename Number = double>

using ryujin::TimeIntegrator< Description, dim, Number >::HyperbolicSystem = typename Description::HyperbolicSystem

Definition at line 260 of file time_integrator.h.

◆ View

template<typename Description , int dim, typename Number = double>

using ryujin::TimeIntegrator< Description, dim, Number >::View = typename Description::template HyperbolicSystemView<dim, Number>

Definition at line 262 of file time_integrator.h.

◆ ParabolicSystem

template<typename Description , int dim, typename Number = double>

using ryujin::TimeIntegrator< Description, dim, Number >::ParabolicSystem = typename Description::ParabolicSystem

Definition at line 265 of file time_integrator.h.

◆ StateVector

template<typename Description , int dim, typename Number = double>

using ryujin::TimeIntegrator< Description, dim, Number >::StateVector = typename View::StateVector

Definition at line 267 of file time_integrator.h.

Constructor & Destructor Documentation

◆ TimeIntegrator()

template<typename Description , int dim, typename Number >

ryujin::TimeIntegrator< Description, dim, Number >::TimeIntegrator	(	const MPIEnsemble &	mpi_ensemble,
		const OfflineData< dim, Number > &	offline_data,
		const HyperbolicModule< Description, dim, Number > &	hyperbolic_module,
		const ParabolicModule< Description, dim, Number > &	parabolic_module,
		const std::string &	subsection = `"/TimeIntegrator< Description, dim, Number >"`
	)

Constructor.

Definition at line 29 of file time_integrator.template.h.

References ryujin::cruise_control, ryujin::erk_33, and ryujin::strang_erk_33_cn.

Member Function Documentation

◆ prepare()

template<typename Description , int dim, typename Number >

void ryujin::TimeIntegrator< Description, dim, Number >::prepare

Prepare time integration. A call to prepare() allocates temporary storage and is necessary before any of the following time-stepping functions can be called.

Definition at line 85 of file time_integrator.template.h.

References ryujin::erk_11, ryujin::erk_22, ryujin::erk_33, ryujin::erk_43, ryujin::erk_54, ryujin::imex_11, ryujin::imex_22, ryujin::imex_33, ryujin::ssprk_22, ryujin::ssprk_33, ryujin::strang_erk_33_cn, ryujin::strang_erk_43_cn, and ryujin::strang_ssprk_33_cn.

◆ prepare_state_vector()

template<typename Description , int dim, typename Number >

void ryujin::TimeIntegrator< Description, dim, Number >::prepare_state_vector	(	StateVector &	state_vector,
		Number	t
	)		const

This function preprocesses a given state vector U for time stepping. It has to be called prior to step() to ensure that precomputed values are in place.

Note: Internally, this function first calls the ParabolicModule::prepare_state_vector() and afterwards HyperbolicModule::prepare_state_vector().

Definition at line 230 of file time_integrator.template.h.

◆ step()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step	(	StateVector &	state_vector,
		Number	t,
		Number	t_final = `std::numeric_limits<Number>::max()`
	)

Given a reference to a previous state vector U performs an explicit time step (and store the result in U). The function returns the chosen time step size tau. The time step size tau is selected such that $t + tau <= t_final$.

Note: This function switches between different Runge-Kutta methods depending on chosen runtime parameters.; Depending on chosen run time parameters different CFL adaptation and recovery strategies for invariant domain violations are used.

Precondition: The state_vector has to be prepared with the prepare_state_vector() function prior to calling the step() function.

Definition at line 247 of file time_integrator.template.h.

References ryujin::erk_11, ryujin::erk_22, ryujin::erk_33, ryujin::erk_43, ryujin::erk_54, ryujin::ssprk_22, ryujin::ssprk_33, and ryujin::strang_ssprk_33_cn.

◆ time_stepping_scheme()

template<typename Description , int dim, typename Number = double>

const auto & ryujin::TimeIntegrator< Description, dim, Number >::time_stepping_scheme ( ) const

inline

The selected time-stepping scheme.

Definition at line 333 of file time_integrator.h.

◆ efficiency()

template<typename Description , int dim, typename Number = double>

const auto & ryujin::TimeIntegrator< Description, dim, Number >::efficiency ( ) const

inline

The eficiency of the selected time-stepping scheme expressed as the ratio of step size of the combined method to step size of an elementary forward Euler step. For example, SSPRK33 has an efficiency ratio of 1 whereas ERK33 has an efficiency ratio of 3.

Definition at line 341 of file time_integrator.h.

◆ step_ssprk_22()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_ssprk_22	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit second-order strong-stability preserving Runge-Kutta SSPRK(2,2;1/2) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 356 of file time_integrator.template.h.

◆ step_ssprk_33()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_ssprk_33	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit third-order strong-stability preserving Runge-Kutta SSPRK(3,3;1/3) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 384 of file time_integrator.template.h.

◆ step_erk_11()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_erk_11	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit first-order Euler step ERK(1,1;1) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 419 of file time_integrator.template.h.

◆ step_erk_22()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_erk_22	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit second-order Runge-Kutta ERK(2,2;1) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 438 of file time_integrator.template.h.

◆ step_erk_33()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_erk_33	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit third-order Runge-Kutta ERK(3,3;1) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 462 of file time_integrator.template.h.

◆ step_erk_43()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_erk_43	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit 4 stage third-order Runge-Kutta ERK(4,3;1) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 497 of file time_integrator.template.h.

◆ step_erk_54()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_erk_54	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an explicit 4 stage fourth-order Runge-Kutta ERK(5,4;1) time step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 537 of file time_integrator.template.h.

◆ step_strang_ssprk_33_cn()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_strang_ssprk_33_cn	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs a combined explicit implicit Strang split using a third-order Runge-Kutta ERK(3,3;1/3) time step and an implicit Crank-Nicolson step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 608 of file time_integrator.template.h.

References ryujin::sadd(), and ryujin::Restart::suggested_tau_max.

◆ step_strang_erk_33_cn()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_strang_erk_33_cn	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs a combined explicit implicit Strang split using a third-order Runge-Kutta ERK(3,3;1) time step and an implicit Crank-Nicolson step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 662 of file time_integrator.template.h.

References ryujin::Restart::suggested_tau_max.

◆ step_strang_erk_43_cn()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_strang_erk_43_cn	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs a combined explicit implicit Strang split using a third-order Runge-Kutta ERK(4,3;1) time step and an implicit Crank-Nicolson step (and store the result in U). The function returns the chosen time step size tau, which is guaranteed to be less than or equal to the parameter tau_max.

Definition at line 723 of file time_integrator.template.h.

References ryujin::Restart::suggested_tau_max.

◆ step_imex_11()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_imex_11	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U, performs an implicit-explicit step IMEX(1,1;1) using a forward euler scheme for the hyperbolic subproblem and backward euler scheme for the parabolic subproblem.

Definition at line 792 of file time_integrator.template.h.

◆ step_imex_22()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_imex_22	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an implicit-explicit IMEX(2,2;1) step using a two stage midpoint rule for the hyperbolic subproblem and a two stage midpoint rule for the parabolic subproblem. The function returns the chosen time step size tau.

Definition at line 815 of file time_integrator.template.h.

◆ step_imex_33()

template<typename Description , int dim, typename Number >

Number ryujin::TimeIntegrator< Description, dim, Number >::step_imex_33	(	StateVector &	state_vector,
		Number	t,
		Number	tau_max
	)

protected

Given a reference to a previous state vector U performs an implicit-explicit IMEX(3,3;1) step using a three stage ERK tableau for the hyperbolic subproblem and a three stage DIRK tableau for the parabolic subproblem. The function returns the chosen time step size tau.

Definition at line 851 of file time_integrator.template.h.

The documentation for this class was generated from the following files:

source/time_integrator.h
source/time_integrator.template.h

Public Types

Public Member Functions

Functions for performing explicit time steps

Detailed Description

Member Typedef Documentation

◆ HyperbolicSystem

◆ View

◆ ParabolicSystem

◆ StateVector

Constructor & Destructor Documentation

◆ TimeIntegrator()

Member Function Documentation

◆ prepare()

◆ prepare_state_vector()

◆ step()

◆ time_stepping_scheme()

◆ efficiency()

◆ step_ssprk_22()

◆ step_ssprk_33()

◆ step_erk_11()

◆ step_erk_22()

◆ step_erk_33()

◆ step_erk_43()

◆ step_erk_54()

◆ step_strang_ssprk_33_cn()

◆ step_strang_erk_33_cn()

◆ step_strang_erk_43_cn()

◆ step_imex_11()

◆ step_imex_22()

◆ step_imex_33()