NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear > Class Reference

Detailed Description

General matrix translator for first order implicit quasi-linear ODEs, used with time discretization schemes that have no special needs.

See also

ODESystemTag::FirstOrderImplicitQuasilinear

Remarks

You might also want read the remarks on time discretization.

Definition at line 90 of file MatrixTranslator.h.

#include <MatrixTranslator.h>

Inheritance diagram for NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >:

Collaboration diagram for NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >:

Public Member Functions
	MatrixTranslatorGeneral (TimeDiscretization const &timeDisc)
void	computeA (GlobalMatrix const &M, GlobalMatrix const &K, GlobalMatrix &A) const override
	Computes \( A = M \cdot \alpha + K \).
void	computeRhs (const GlobalMatrix &M, const GlobalMatrix &, const GlobalVector &b, const GlobalVector &x_prev, GlobalVector &rhs) const override
	Computes \( \mathtt{rhs} = M \cdot x_O + b \).
void	normalizeAandRhs (GlobalMatrix &A, GlobalVector &b) const override
void	computeResidual (GlobalMatrix const &M, GlobalMatrix const &K, GlobalVector const &b, double dt, GlobalVector const &x_curr, GlobalVector const &x_prev, GlobalVector &res) const override
	Computes \( r = M \cdot \hat x + K \cdot x_C - b \).
void	computeJacobian (GlobalMatrix const &Jac_in, GlobalMatrix &Jac_out) const override
Public Member Functions inherited from NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >
virtual	~MatrixTranslator ()=default

Private Attributes
TimeDiscretization const &	_time_disc
	the time discretization used.
std::size_t	_tmp_id = 0u
	ID of the vector storing intermediate computations.

Constructor & Destructor Documentation

MatrixTranslatorGeneral()

NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::MatrixTranslatorGeneral ( TimeDiscretization const & timeDisc )

inlineexplicit

Constructs a new instance.

Parameters

timeDisc

the time discretization scheme to be used.

Definition at line 98 of file MatrixTranslator.h.

        : _time_disc(timeDisc)
    {
    }

Member Function Documentation

computeA()

void NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeA ( GlobalMatrix const & M, GlobalMatrix const & K, GlobalMatrix & A ) const

overridevirtual

Computes \( A = M \cdot \alpha + K \).

Implements NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >.

Definition at line 17 of file MatrixTranslator.cpp.

{
    namespace LinAlg = MathLib::LinAlg;
 
    double const dt = _time_disc.getCurrentTimeIncrement();
 
    // A = M * 1/dt + K
    LinAlg::copy(K, A);
    LinAlg::axpy(A, 1. / dt, M);
}

References MathLib::LinAlg::axpy(), and MathLib::LinAlg::copy().

computeJacobian()

void NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeJacobian ( GlobalMatrix const & Jac_in, GlobalMatrix & Jac_out ) const

overridevirtual

Writes Jac_in to Jac_out.

Todo

Do not copy.

Implements NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >.

Definition at line 83 of file MatrixTranslator.cpp.

{
    namespace LinAlg = MathLib::LinAlg;
 
    LinAlg::copy(Jac_in, Jac_out);
}

References MathLib::LinAlg::copy().

computeResidual()

void NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeResidual ( GlobalMatrix const & M, GlobalMatrix const & K, GlobalVector const & b, double dt, GlobalVector const & x_curr, GlobalVector const & x_prev, GlobalVector & res ) const

overridevirtual

Computes \( r = M \cdot \hat x + K \cdot x_C - b \).

Implements NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >.

Definition at line 65 of file MatrixTranslator.cpp.

{
    namespace LinAlg = MathLib::LinAlg;
 
    // res = M * x_dot + K * x_curr - b
    GlobalVector x_dot;
    LinAlg::copy(x_curr, x_dot);              // x_dot = x
    LinAlg::axpy(x_dot, -1., x_prev);         // x_dot = x - x_prev
    LinAlg::scale(x_dot, 1. / dt);            // x_dot = (x - x_prev)/dt
    LinAlg::matMult(M, x_dot, res);           // res = M*x_dot
    LinAlg::matMultAdd(K, x_curr, res, res);  // res = M*x_dot + K*x
    LinAlg::axpy(res, -1., b);                // res = M*x_dot + X*x - b
}

References MathLib::LinAlg::axpy(), MathLib::LinAlg::copy(), MathLib::LinAlg::matMult(), MathLib::LinAlg::matMultAdd(), and MathLib::LinAlg::scale().

computeRhs()

void NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeRhs ( const GlobalMatrix & M, const GlobalMatrix & , const GlobalVector & b, const GlobalVector & x_prev, GlobalVector & rhs ) const

overridevirtual

Computes \( \mathtt{rhs} = M \cdot x_O + b \).

Implements NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >.

Definition at line 30 of file MatrixTranslator.cpp.

{
    namespace LinAlg = MathLib::LinAlg;
 
    auto& tmp = NumLib::GlobalVectorProvider::provider.getVector(_tmp_id);
    _time_disc.getWeightedOldX(tmp, x_prev);
 
    // rhs = M * weighted_old_x + b
    LinAlg::matMultAdd(M, tmp, b, rhs);
 
    NumLib::GlobalVectorProvider::provider.releaseVector(tmp);
}

References NumLib::VectorProvider::getVector(), MathLib::LinAlg::matMultAdd(), NumLib::GlobalVectorProvider::provider, and NumLib::VectorProvider::releaseVector().

normalizeAandRhs()

void NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::normalizeAandRhs ( GlobalMatrix & A, GlobalVector & b ) const

overridevirtual

Computes \( A = A^T \cdot A \), and also \( rhs = A^T \cdot rhs \).

Implements NumLib::MatrixTranslator< ODESystemTag::FirstOrderImplicitQuasilinear >.

Definition at line 46 of file MatrixTranslator.cpp.

{
    namespace LinAlg = MathLib::LinAlg;
 
    // check whether A is square?
 
    GlobalMatrix new_A(A);
    GlobalVector new_b(b);
    LinAlg::copy(A, new_A);
    LinAlg::copy(b, new_b);
    // rhs = A^T * rhs
    // A = A^T * A
    LinAlg::linearSysNormalize(A, new_A, b, new_b);
 
    LinAlg::copy(new_A, A);
    LinAlg::copy(new_b, b);
}

References MathLib::LinAlg::copy(), and MathLib::LinAlg::linearSysNormalize().

Member Data Documentation

_time_disc

TimeDiscretization const& NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::_time_disc

private

the time discretization used.

Definition at line 128 of file MatrixTranslator.h.

_tmp_id

std::size_t NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::_tmp_id = 0u

mutableprivate

ID of the vector storing intermediate computations.

Definition at line 131 of file MatrixTranslator.h.

---

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

• NumLib/ODESolver/MatrixTranslator.h
• NumLib/ODESolver/MatrixTranslator.cpp