Class CIBSaddlePointSolver solves for the fluid velocity $\vec{v}$ , fluid pressure $ p $ , and the Lagrange multiplier $\vec{\lambda}$ maintaining the rigidity constraint. More...

#include </home/runner/work/IBAMR/IBAMR/include/ibamr/CIBSaddlePointSolver.h>

Inheritance diagram for IBAMR::CIBSaddlePointSolver:

[legend]

Public Member Functions
	CIBSaddlePointSolver (std::string object_name, SAMRAI::tbox::Pointer< SAMRAI::tbox::Database > input_db, SAMRAI::tbox::Pointer< IBAMR::INSStaggeredHierarchyIntegrator > navier_stokes_integrator, SAMRAI::tbox::Pointer< IBAMR::CIBStrategy > cib_strategy, std::string default_options_prefix, MPI_Comm petsc_comm=PETSC_COMM_WORLD)
	Constructor for a saddle-point solver that employs the PETSc KSP solver framework.

virtual	~CIBSaddlePointSolver ()
	Destructor.

void	setKSPType (const std::string &ksp_type)
	Set the KSP type.

void	setOptionsPrefix (const std::string &options_prefix)
	Set the options prefix used by this PETSc solver object.

void	setVelocityPoissonSpecifications (const SAMRAI::solv::PoissonSpecifications &u_problem_coefs)
	Set the PoissonSpecifications object used to specify the coefficients for the momentum equation in the incompressible Stokes operator.

void	setPhysicalBcCoefs (const std::vector< SAMRAI::solv::RobinBcCoefStrategy< NDIM > * > &u_bc_coefs, SAMRAI::solv::RobinBcCoefStrategy< NDIM > *p_bc_coef)
	Set the SAMRAI::solv::RobinBcCoefStrategy objects used to specify physical boundary conditions. More...

void	setPhysicalBoundaryHelper (SAMRAI::tbox::Pointer< IBAMR::StaggeredStokesPhysicalBoundaryHelper > bc_helper)
	Set the StokesSpecifications object and timestep size used to specify the coefficients for the time-dependent incompressible Stokes operator.

SAMRAI::tbox::Pointer< IBTK::LinearOperator >	getA () const
	Return the linear operator for the saddle-point solver.

SAMRAI::tbox::Pointer< IBAMR::StaggeredStokesSolver >	getStokesSolver () const
	Return the Stokes solver used in the preconditioner of the solver.

double	getInterpScale () const
	Return the interpolation scale factor.

double	getSpreadScale () const
	Return the spreading scale factor.

double	getRegularizationWeight () const
	Return the regularization scale factor.

bool	getNormalizeSpreadForce () const
	Return if the spread force is to be normalized.

bool	solveSystem (Vec x, Vec b)
	Solve the linear system of equations for . More...

void	initializeSolverState (Vec x, Vec b)
	Compute hierarchy dependent data required for solving .

void	deallocateSolverState ()
	Remove all hierarchy dependent data allocated by initializeSolverState(). More...

void	setSolutionTime (double solution_time)
	Set the time at which the solution is to be evaluated.

void	setTimeInterval (double current_time, double new_time)
	Set the current time interval.

SAMRAI::tbox::Pointer< IBAMR::CIBMobilitySolver >	getCIBMobilitySolver () const
	Get the mobility solver.

Detailed Description

Class CIBSaddlePointSolver solves for the fluid velocity $\vec{v}$ , fluid pressure $ p $ , and the Lagrange multiplier $\vec{\lambda}$ maintaining the rigidity constraint.

For free-moving (self-moving bodies) it also solves for the rigid body translational and rotational velocities $\vec{U}$ .

For free-moving case, the class solves a saddle-point problem of the type:

$\begin{eqnarray*} A \vec{v} + \nabla p - \gamma S \vec{\lambda} &=& \vec{g} \\ -\nabla \cdot \vec{v} &=& (h = 0) \\ -\beta J \vec{v} + \beta T^{*} \vec{U} -\beta \delta \vec{\lambda} &=& (\vec{w} = -\beta \vec{U}_{\mbox{\scriptsize def}}) \\ T \vec{\lambda} &=& \vec{F} \end{eqnarray*}$

For imposed kinematics case, the class solves:

$\begin{eqnarray*} A \vec{v} + \nabla p - \gamma S \vec{\lambda} &=& \vec{g} \\ -\nabla \cdot \vec{v} &=& (h = 0) \\ -\beta J \vec{v} - \beta \delta \vec{\lambda} &=& (w = -\beta (T^{*}\vec{U}+ \vec{U}_{\mbox{\scriptsize def}}) = -\beta \vec{U}_{\mbox{\scriptsize imposed}}) \end{eqnarray*}$

Here, $ A $ is the Helmholtz operator, $ S $ is the spreading operator, $ G $ is the gradient operator, $ J $ is the interpolation operator, and $ T $ is the rigid body operator (with $T^{*}$ denoting its adjoint), $\vec{F}$ is the net external force/torque on the body (excluding the hydrodynamic force/torque), $\vec{U}_{\mbox{\scriptsize def}}$ is the deformational velocity of the body, and $\vec{U}_{\mbox{\scriptsize imposed}}$ is the velocity imposed on the body. $\gamma$ and $\beta$ are two scaling parameters to make the system well-scaled. $\delta$ is the regularization parameter that regularizes the problem in presence of many IB/control points in a grid cell.

Here, we employ the Krylov solver to solve the above saddle-point problem. We use Schur complement preconditioner to precondition the iterative solver.

Member Function Documentation

◆ deallocateSolverState()

void IBAMR::CIBSaddlePointSolver::deallocateSolverState ( )

Remove all hierarchy dependent data allocated by initializeSolverState().

Note: It is safe to call deallocateSolverState() when the solver state is already deallocated.

See also: initializeSolverState

◆ setPhysicalBcCoefs()

void IBAMR::CIBSaddlePointSolver::setPhysicalBcCoefs	(	const std::vector< SAMRAI::solv::RobinBcCoefStrategy< NDIM > * > &	u_bc_coefs,
		SAMRAI::solv::RobinBcCoefStrategy< NDIM > *	p_bc_coef
	)

Set the SAMRAI::solv::RobinBcCoefStrategy objects used to specify physical boundary conditions.

Note: Any of the elements of u_bc_coefs may be NULL. In this case, homogeneous Dirichlet boundary conditions are employed for that data depth. p_bc_coef may also be NULL; in that case, homogeneous Neumann boundary conditions are employed for the pressure.

Parameters

u_bc_coefs	IBTK::Vector of pointers to objects that can set the Robin boundary condition coefficients for the velocity.
p_bc_coef	Pointer to object that can set the Robin boundary condition coefficients for the pressure.

◆ solveSystem()

bool IBAMR::CIBSaddlePointSolver::solveSystem	(	Vec	x,
		Vec	b
	)

Solve the linear system of equations $Ax=b$ for $x$ .

Returns: true if the solver converged to the specified tolerances, false otherwise

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

/home/runner/work/IBAMR/IBAMR/include/ibamr/CIBSaddlePointSolver.h
/home/runner/work/IBAMR/IBAMR/src/IB/CIBSaddlePointSolver.cpp

Public Member Functions

Detailed Description

Member Function Documentation

◆ deallocateSolverState()

◆ setPhysicalBcCoefs()

◆ solveSystem()