Skip to content

Class ae108::cpppetsc::TAOSolver

template <class Policy>

ClassList > ae108 > cpppetsc > TAOSolver

Classes

Type Name
struct EqualityConstraints

Public Types

Type Name
typedef std::function< void(const distributed< vector_type > &, distributed< vector_type > *)> GradientFunctor
typedef std::function< void(const distributed< vector_type > &, matrix_type *)> HessianFunctor
typedef std::function< void(const distributed< vector_type > &, real_type *)> ObjectiveFunctor
enum std::uint8_t Type
typedef Matrix< Policy > matrix_type
typedef typename Vector< Policy >::real_type real_type
typedef typename Vector< Policy >::size_type size_type
typedef typename Vector< Policy >::value_type value_type
typedef Vector< Policy > vector_type

Public Static Attributes

Type Name
constexpr value_type no_lower_bound = PETSC\_NINFINITY
constexpr value_type no_upper_bound = PETSC\_INFINITY

Public Functions

Type Name
TAOSolver (matrix_type bufferMatrix)
Initializes the nonlinear solver with the problem dimension.
TAOSolver (matrix_type bufferMatrix, const Type defaultType)
Initializes the nonlinear solver with the problem dimension.
void setBounds (const distributed< vector_type > & lowerBounds, const distributed< vector_type > & upperBounds)
Define upper and lower bounds for the variables.
void setConstraints (EqualityConstraints constraints)
Set the equality constraints of the problem.
void setType (const Type type)
Set the type of the optimizer.
distributed< vector_type > solve (ObjectiveFunctor objective, GradientFunctor gradient, HessianFunctor hessian, distributed< vector_type > guess) const
Minimizes the objective function using optimization.

Public Types Documentation

typedef GradientFunctor

using ae108::cpppetsc::TAOSolver< Policy >::GradientFunctor =  std::function<void(const distributed<vector_type> &, distributed<vector_type> *)>;

Remark:

The first parameter is the input, the second parameter is the output.

typedef HessianFunctor

using ae108::cpppetsc::TAOSolver< Policy >::HessianFunctor =  std::function<void(const distributed<vector_type> &, matrix_type *)>;

Remark:

The first parameter is the input, the second parameter is the output.

typedef ObjectiveFunctor

using ae108::cpppetsc::TAOSolver< Policy >::ObjectiveFunctor =  std::function<void(const distributed<vector_type> &, real_type *)>;

Remark:

The first parameter is the input, the second parameter is the output.

enum Type

enum ae108::cpppetsc::TAOSolver::Type {
    lmvm,
    nls,
    ntr,
    ntl,
    cg,
    tron,
    owlqn,
    bmrm,
    blmvm,
    bqnls,
    bncg,
    bnls,
    bntr,
    bqnkls,
    bqnktr,
    bqnktl,
    bqpip,
    gpcg,
    nm,
    pounders,
    brgn,
    lcl,
    ssils,
    ssfls,
    asils,
    asfls,
    ipm,
    pdipm,
    shell,
    admm,
    almm
};

typedef matrix_type

using ae108::cpppetsc::TAOSolver< Policy >::matrix_type =  Matrix<Policy>;

typedef real_type

using ae108::cpppetsc::TAOSolver< Policy >::real_type =  typename Vector<Policy>::real_type;

typedef size_type

using ae108::cpppetsc::TAOSolver< Policy >::size_type =  typename Vector<Policy>::size_type;

typedef value_type

using ae108::cpppetsc::TAOSolver< Policy >::value_type =  typename Vector<Policy>::value_type;

typedef vector_type

using ae108::cpppetsc::TAOSolver< Policy >::vector_type =  Vector<Policy>;

Public Static Attributes Documentation

variable no_lower_bound

constexpr value_type ae108::cpppetsc::TAOSolver< Policy >::no_lower_bound;

variable no_upper_bound

constexpr value_type ae108::cpppetsc::TAOSolver< Policy >::no_upper_bound;

Public Functions Documentation

function TAOSolver [1/2]

Initializes the nonlinear solver with the problem dimension. 

explicit ae108::cpppetsc::TAOSolver::TAOSolver (
    matrix_type bufferMatrix
)

Parameters:

  • The matrix to use as an internal buffer.

function TAOSolver [2/2]

Initializes the nonlinear solver with the problem dimension. 

explicit ae108::cpppetsc::TAOSolver::TAOSolver (
    matrix_type bufferMatrix,
    const Type defaultType
)

Parameters:

  • The matrix to use as an internal buffer.
  • defaultType The default type of solver used. This type will not be used if e.g. a command line option specifying the type is present (see PETSc documentation).

function setBounds

Define upper and lower bounds for the variables. 

void ae108::cpppetsc::TAOSolver::setBounds (
    const distributed < vector_type > & lowerBounds,
    const distributed < vector_type > & upperBounds
)

Parameters:

  • lowerBounds A vector of constraints for every variable. Use no_lower_bound to set no constraints for a variable.
  • upperBounds A vector of constraints for every variable. Use no_upper_bound to set no constraints for a variable.

function setConstraints

Set the equality constraints of the problem. 

void ae108::cpppetsc::TAOSolver::setConstraints (
    EqualityConstraints constraints
)

Note:

Not all solvers consider these constraints.

function setType

Set the type of the optimizer. 

void ae108::cpppetsc::TAOSolver::setType (
    const Type type
)

Parameters:

  • type The type of the optimizer. See the definition of Type and the PETSc documentation for more information about the possible choices.

function solve

Minimizes the objective function using optimization. 

distributed < vector_type > ae108::cpppetsc::TAOSolver::solve (
    ObjectiveFunctor objective,
    GradientFunctor gradient,
    HessianFunctor hessian,
    distributed < vector_type > guess
) const

Parameters:

  • objective The function to minimize.
  • gradient The gradient of the objective.
  • hessian The hessian of the objective.
  • guess The starting point for the optimization.

Exception:

Returns:

The minimizer.

The documentation for this class was generated from the following file cpppetsc/src/include/ae108/cpppetsc/TAOSolver.h