#include <alpaqa/problem/type-erased-problem.hpp>

Detailed Description

template<Config Conf = DefaultConfig, class Allocator = std::allocator<std::byte>>
class alpaqa::TypeErasedProblem< Conf, Allocator >

The main polymorphic minimization problem interface.

This class wraps the actual problem implementation class, filling in the missing member functions with sensible defaults, and providing a uniform interface that is used by the solvers.

The problem implementations do not inherit from an abstract base class. Instead, structural typing is used. The ProblemVTable constructor uses reflection to discover which member functions are provided by the problem implementation. See Problem formulations for more information, and C++/CustomCppProblem/main.cpp for an example.

Definition at line 222 of file type-erased-problem.hpp.

Inheritance diagram for TypeErasedProblem< Conf, Allocator >:

Collaboration diagram for TypeErasedProblem< Conf, Allocator >:

Problem dimensions
length_t	get_n () const
	[Required] Number of decision variables.

length_t	get_m () const
	[Required] Number of constraints.

Required cost and constraint functions
real_t	eval_f (crvec x) const
	[Required] Function that evaluates the cost, \( f(x) \)

void	eval_grad_f (crvec x, rvec grad_fx) const
	[Required] Function that evaluates the gradient of the cost, \( \nabla f(x) \)

void	eval_g (crvec x, rvec gx) const
	[Required] Function that evaluates the constraints, \( g(x) \)

void	eval_grad_g_prod (crvec x, crvec y, rvec grad_gxy) const
	[Required] Function that evaluates the gradient of the constraints times a vector, \( \nabla g(x)\,y = \tp{\jac_g(x)}y \)

Projections onto constraint sets and proximal mappings
void	eval_proj_diff_g (crvec z, rvec e) const
	[Required] Function that evaluates the difference between the given point \( z \) and its projection onto the constraint set \( D \).

void	eval_proj_multipliers (rvec y, real_t M) const
	[Required] Function that projects the Lagrange multipliers for ALM.

real_t	eval_prox_grad_step (real_t γ, crvec x, crvec grad_ψ, rvec x̂, rvec p) const
	[Required] Function that computes a proximal gradient step.

index_t	eval_inactive_indices_res_lna (real_t γ, crvec x, crvec grad_ψ, rindexvec J) const
	[Optional] Function that computes the inactive indices \( \mathcal J(x) \) for the evaluation of the linear Newton approximation of the residual, as in [2].

Constraint sets
const Box &	get_box_C () const
	[Optional] Get the rectangular constraint set of the decision variables, \( x \in C \).

const Box &	get_box_D () const
	[Optional] Get the rectangular constraint set of the general constraint function, \( g(x) \in D \).

Functions for second-order solvers
void	eval_jac_g (crvec x, rindexvec inner_idx, rindexvec outer_ptr, rvec J_values) const
	[Optional] Function that evaluates the Jacobian of the constraints as a sparse matrix, \( \jac_g(x) \)

length_t	get_jac_g_num_nonzeros () const
	[Optional] Function that gets the number of nonzeros of the sparse Jacobian of the constraints.

void	eval_grad_gi (crvec x, index_t i, rvec grad_gi) const
	[Optional] Function that evaluates the gradient of one specific constraint, \( \nabla g_i(x) \)

void	eval_hess_L_prod (crvec x, crvec y, real_t scale, crvec v, rvec Hv) const
	[Optional] Function that evaluates the Hessian of the Lagrangian multiplied by a vector, \( \nabla_{xx}^2L(x, y)\,v \)

void	eval_hess_L (crvec x, crvec y, real_t scale, rindexvec inner_idx, rindexvec outer_ptr, rvec H_values) const
	[Optional] Function that evaluates the Hessian of the Lagrangian as a sparse matrix, \( \nabla_{xx}^2L(x, y) \)

length_t	get_hess_L_num_nonzeros () const
	[Optional] Function that gets the number of nonzeros of the sparse Hessian of the Lagrangian.

void	eval_hess_ψ_prod (crvec x, crvec y, crvec Σ, real_t scale, crvec v, rvec Hv) const
	[Optional] Function that evaluates the Hessian of the augmented Lagrangian multiplied by a vector, \( \nabla_{xx}^2L_\Sigma(x, y)\,v \)

void	eval_hess_ψ (crvec x, crvec y, crvec Σ, real_t scale, rindexvec inner_idx, rindexvec outer_ptr, rvec H_values) const
	[Optional] Function that evaluates the Hessian of the augmented Lagrangian, \( \nabla_{xx}^2L_\Sigma(x, y) \)

length_t	get_hess_ψ_num_nonzeros () const
	[Optional] Function that gets the number of nonzeros of the Hessian of the augmented Lagrangian.

Combined evaluations
real_t	eval_f_grad_f (crvec x, rvec grad_fx) const
	[Optional] Evaluate both \( f(x) \) and its gradient, \( \nabla f(x) \).

real_t	eval_f_g (crvec x, rvec g) const
	[Optional] Evaluate both \( f(x) \) and \( g(x) \).

void	eval_grad_f_grad_g_prod (crvec x, crvec y, rvec grad_f, rvec grad_gxy) const
	[Optional] Evaluate both \( \nabla f(x) \) and \( \nabla g(x)\,y \).

void	eval_grad_L (crvec x, crvec y, rvec grad_L, rvec work_n) const
	[Optional] Evaluate the gradient of the Lagrangian \( \nabla_x L(x, y) = \nabla f(x) + \nabla g(x)\,y \)

Augmented Lagrangian
real_t	eval_ψ (crvec x, crvec y, crvec Σ, rvec ŷ) const
	[Optional] Calculate both ψ(x) and the vector ŷ that can later be used to compute ∇ψ.

void	eval_grad_ψ (crvec x, crvec y, crvec Σ, rvec grad_ψ, rvec work_n, rvec work_m) const
	[Optional] Calculate the gradient ∇ψ(x).

real_t	eval_ψ_grad_ψ (crvec x, crvec y, crvec Σ, rvec grad_ψ, rvec work_n, rvec work_m) const
	[Optional] Calculate both ψ(x) and its gradient ∇ψ(x).

Checks
void	check () const
	[Optional] Check that the problem formulation is well-defined, the dimensions match, etc.

Querying specialized implementations
bool	provides_eval_inactive_indices_res_lna () const
	Returns true if the problem provides an implementation of eval_inactive_indices_res_lna.

bool	provides_eval_jac_g () const
	Returns true if the problem provides an implementation of eval_jac_g.

bool	provides_get_jac_g_num_nonzeros () const
	Returns true if the problem provides an implementation of get_jac_g_num_nonzeros.

bool	provides_eval_grad_gi () const
	Returns true if the problem provides an implementation of eval_grad_gi.

bool	provides_eval_hess_L_prod () const
	Returns true if the problem provides an implementation of eval_hess_L_prod.

bool	provides_eval_hess_L () const
	Returns true if the problem provides an implementation of eval_hess_L.

bool	provides_get_hess_L_num_nonzeros () const
	Returns true if the problem provides an implementation of get_hess_L_num_nonzeros.

bool	provides_eval_hess_ψ_prod () const
	Returns true if the problem provides an implementation of eval_hess_ψ_prod.

bool	provides_eval_hess_ψ () const
	Returns true if the problem provides an implementation of eval_hess_ψ.

bool	provides_get_hess_ψ_num_nonzeros () const
	Returns true if the problem provides an implementation of get_hess_ψ_num_nonzeros.

bool	provides_eval_f_grad_f () const
	Returns true if the problem provides a specialized implementation of eval_f_grad_f, false if it uses the default implementation.

bool	provides_eval_f_g () const
	Returns true if the problem provides a specialized implementation of eval_f_g, false if it uses the default implementation.

bool	provides_eval_grad_f_grad_g_prod () const
	Returns true if the problem provides a specialized implementation of eval_grad_f_grad_g_prod, false if it uses the default implementation.

bool	provides_eval_grad_L () const
	Returns true if the problem provides a specialized implementation of eval_grad_L, false if it uses the default implementation.

bool	provides_eval_ψ () const
	Returns true if the problem provides a specialized implementation of eval_ψ, false if it uses the default implementation.

bool	provides_eval_grad_ψ () const
	Returns true if the problem provides a specialized implementation of eval_grad_ψ, false if it uses the default implementation.

bool	provides_eval_ψ_grad_ψ () const
	Returns true if the problem provides a specialized implementation of eval_ψ_grad_ψ, false if it uses the default implementation.

bool	provides_get_box_C () const
	Returns true if the problem provides an implementation of get_box_C.

bool	provides_get_box_D () const
	Returns true if the problem provides an implementation of get_box_D.

bool	provides_check () const
	Returns true if the problem provides an implementation of check.

Helpers
real_t	calc_ŷ_dᵀŷ (rvec g_ŷ, crvec y, crvec Σ) const
	Given g(x), compute the intermediate results ŷ and dᵀŷ that can later be used to compute ψ(x) and ∇ψ(x).

Public Types
using	Box = alpaqa::Box< config_t >

using	VTable = ProblemVTable< config_t >

using	allocator_type = Allocator

using	TypeErased = util::TypeErased< VTable, allocator_type >

Public Member Functions
	TypeErased () noexcept(noexcept(allocator_type()))=default
	Default constructor.

template<class Alloc >
	TypeErased (std::allocator_arg_t, const Alloc &alloc)
	Default constructor (allocator aware).

	TypeErased (const TypeErased &other)
	Copy constructor.

	TypeErased (const TypeErased &other, allocator_type alloc)
	Copy constructor (allocator aware).

	TypeErased (TypeErased &&other) noexcept
	Move constructor.

	TypeErased (TypeErased &&other, const allocator_type &alloc) noexcept
	Move constructor (allocator aware).

template<class T , class Alloc >
	TypeErased (std::allocator_arg_t, const Alloc &alloc, T &&d)
	Main constructor that type-erases the given argument.

template<class T , class Alloc , class... Args>
	TypeErased (std::allocator_arg_t, const Alloc &alloc, te_in_place_t< T >, Args &&...args)
	Main constructor that type-erases the object constructed from the given argument.

template<class T > requires no_child_of_ours<T>
	TypeErased (T &&d)

template<class T , class... Args>
	TypeErased (te_in_place_t< T >, Args &&...args)
	Main constructor that type-erases the object constructed from the given argument.

	operator bool () const noexcept
	Check if this wrapper wraps an object.

allocator_type	get_allocator () const noexcept
	Get a copy of the allocator.

const std::type_info &	type () const noexcept
	Query the contained type.

template<class T >
T &	as () &
	Convert the type-erased object to the given type.

template<class T >
const T &	as () const &
	Convert the type-erased object to the given type.

template<class T >
const T &&	as () &&
	Convert the type-erased object to the given type.

Static Public Member Functions
template<class T , class... Args>
static TypeErasedProblem	make (Args &&...args)

template<class Ret , class T , class Alloc , class... Args> requires std::is_base_of_v<TypeErased, Ret>
static Ret	make (std::allocator_arg_t tag, const Alloc &alloc, Args &&...args)
	Construct a type-erased wrapper of type Ret for an object of type T, initialized in-place with the given arguments.

Static Public Attributes
static constexpr size_t	small_buffer_size = SmallBufferSize

Protected Member Functions
template<class Ret , class... FArgs, class... Args>
decltype(auto)	call (Ret(f)(const void , FArgs...), Args &&...args) const
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret , class... FArgs, class... Args>
decltype(auto)	call (Ret(f)(void , FArgs...), Args &&...args)
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret >
decltype(auto)	call (Ret(f)(const void )) const
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret >
decltype(auto)	call (Ret(f)(void ))
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret >
decltype(auto)	call (Ret(f)(const void , const VTable &)) const
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret >
decltype(auto)	call (Ret(f)(void , const VTable &))
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class T , class... Args>
void	construct_inplace (Args &&...args)
	Ensure storage and construct the type-erased object of type T in-place.

template<class Ret >
decltype(auto)	call (Ret(f)(const void , const VTable &)) const
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

template<class Ret >
decltype(auto)	call (Ret(f)(void , const VTable &))
	Call the vtable function `f` with the given arguments `args`, implicitly passing the self pointer and vtable reference if necessary.

Protected Attributes
void *	self
	Pointer to the stored object.

VTable	vtable

size_t	size = invalid_size
	Size required to store the object.

Static Protected Attributes
static constexpr size_t	invalid_size

Private Types
using	allocator_traits = std::allocator_traits< allocator_type >

using	buffer_type = std::array< std::byte, small_buffer_size >

Private Member Functions
Deallocator	allocate (size_t size)
	Ensure that storage is available, either by using the small buffer if it is large enough, or by calling the allocator.

void	deallocate ()
	Deallocate the memory without invoking the destructor.

void	cleanup ()
	Destroy the type-erased object (if not empty), and deallocate the memory if necessary.

template<bool CopyAllocator>
void	do_copy_assign (const TypeErased &other)

Private Attributes
buffer_type	small_buffer

allocator_type	allocator

Static Private Attributes
template<class T >
static constexpr auto	no_child_of_ours
	True if `T` is not a child class of TypeErased.

Member Typedef Documentation

◆ Box

using Box = alpaqa::Box<config_t>

Definition at line 225 of file type-erased-problem.hpp.

◆ VTable

using VTable = ProblemVTable<config_t>

Definition at line 226 of file type-erased-problem.hpp.

◆ allocator_type

using allocator_type = Allocator

Definition at line 227 of file type-erased-problem.hpp.

◆ TypeErased

using TypeErased = util::TypeErased<VTable, allocator_type>

Definition at line 228 of file type-erased-problem.hpp.

◆ allocator_traits

using allocator_traits = std::allocator_traits<allocator_type>

privateinherited

Definition at line 200 of file type-erasure.hpp.

◆ buffer_type

using buffer_type = std::array<std::byte, small_buffer_size>

privateinherited

Definition at line 201 of file type-erasure.hpp.

Member Function Documentation

◆ make() [1/2]

static TypeErasedProblem make ( Args &&... args )

inlinestatic

Definition at line 238 of file type-erased-problem.hpp.

◆ get_n()

auto get_n

[Required] Number of decision variables.

Definition at line 727 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_m()

auto get_m

[Required] Number of constraints.

Definition at line 731 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_f()

auto eval_f ( crvec x ) const

[Required] Function that evaluates the cost, \( f(x) \)

Parameters

[in] x Decision variable \( x \in \R^n \)

Definition at line 756 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_grad_f()

void eval_grad_f	(	crvec	x,
		rvec	grad_fx
	)		const

[Required] Function that evaluates the gradient of the cost, \( \nabla f(x) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[out]	grad_fx	Gradient of cost function \( \nabla f(x) \in \R^n \)

Definition at line 760 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_g()

void eval_g	(	crvec	x,
		rvec	gx
	)		const

[Required] Function that evaluates the constraints, \( g(x) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[out]	gx	Value of the constraints \( g(x) \in \R^m \)

Definition at line 764 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_grad_g_prod()

void eval_grad_g_prod	(	crvec	x,
		crvec	y,
		rvec	grad_gxy
	)		const

[Required] Function that evaluates the gradient of the constraints times a vector, \( \nabla g(x)\,y = \tp{\jac_g(x)}y \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	y	Vector \( y \in \R^m \) to multiply the gradient by
[out]	grad_gxy	Gradient of the constraints \( \nabla g(x)\,y \in \R^n \)

Definition at line 768 of file type-erased-problem.hpp.

◆ eval_proj_diff_g()

void eval_proj_diff_g	(	crvec	z,
		rvec	e
	)		const

[Required] Function that evaluates the difference between the given point \( z \) and its projection onto the constraint set \( D \).

Parameters

[in]	z	Slack variable, \( z \in \R^m \)
[out]	e	The difference relative to its projection, \( e = z - \Pi_D(z) \in \R^m \)

Note: z and e can refer to the same vector.

Definition at line 736 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_proj_multipliers()

void eval_proj_multipliers	(	rvec	y,
		real_t	M
	)		const

[Required] Function that projects the Lagrange multipliers for ALM.

Parameters

[in,out]	y	Multipliers, \( y \leftarrow \Pi_Y(y) \in \R^m \)
[in]	M	The radius/size of the set \( Y \). See ALMParams::max_multiplier.

Definition at line 740 of file type-erased-problem.hpp.

◆ eval_prox_grad_step()

auto eval_prox_grad_step	(	real_t	γ,
		crvec	x,
		crvec	grad_ψ,
		rvec	x̂,
		rvec	p
	)		const

[Required] Function that computes a proximal gradient step.

Parameters

[in]	γ	Step size, \( \gamma \in \R_{>0} \)
[in]	x	Decision variable \( x \in \R^n \)
[in]	grad_ψ	Gradient of the subproblem cost, \( \nabla\psi(x) \in \R^n \)
[out]	x̂	Next proximal gradient iterate, \( \hat x = T_\gamma(x) = \prox_{\gamma h}(x - \gamma\nabla\psi(x)) \in \R^n \)
[out]	p	The proximal gradient step, \( p = \hat x - x \in \R^n \)

Returns: The nonsmooth function evaluated at x̂, \( h(\hat x) \).

Note: The vector \( p \) is often used in stopping criteria, so its numerical accuracy is more important than that of \( \hat x \).

Definition at line 744 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_inactive_indices_res_lna()

auto eval_inactive_indices_res_lna	(	real_t	γ,
		crvec	x,
		crvec	grad_ψ,
		rindexvec	J
	)		const

[Optional] Function that computes the inactive indices \( \mathcal J(x) \) for the evaluation of the linear Newton approximation of the residual, as in [2].

Parameters

[in]	γ	Step size, \( \gamma \in \R_{>0} \)
[in]	x	Decision variable \( x \in \R^n \)
[in]	grad_ψ	Gradient of the subproblem cost, \( \nabla\psi(x) \in \R^n \)
[out]	J	The indices of the components of \( x \) that are in the index set \( \mathcal J(x) \). In ascending order, at most n.

Returns: The number of inactive constraints, \( \# \mathcal J(x) \).

For example, in the case of box constraints, we have

\[ \mathcal J(x) \defeq \defset{i \in \N_{[0, n-1]}}{\underline x_i \lt x_i - \gamma\nabla_{\!x_i}\psi(x) \lt \overline x_i}. \]

Definition at line 749 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_box_C()

auto get_box_C

[Optional] Get the rectangular constraint set of the decision variables, \( x \in C \).

Definition at line 851 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_box_D()

auto get_box_D

[Optional] Get the rectangular constraint set of the general constraint function, \( g(x) \in D \).

Definition at line 855 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_jac_g()

void eval_jac_g	(	crvec	x,
		rindexvec	inner_idx,
		rindexvec	outer_ptr,
		rvec	J_values
	)		const

[Optional] Function that evaluates the Jacobian of the constraints as a sparse matrix, \( \jac_g(x) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in,out]	inner_idx	Inner indices (row indices of nonzeros).
[in,out]	outer_ptr	Outer pointers (points to the first nonzero in each column).
[out]	J_values	Nonzero values of the Jacobian \( \jac_g(x) \in \R^{m\times n} \) If `J_values` has size zero, this function should initialize `inner_idx` and `outer_ptr`. If `J_values` is nonempty, `inner_idx` and `outer_ptr` can be assumed to be initialized, and this function should evaluate `J_values`.

Required for second-order solvers only.

Definition at line 776 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_jac_g_num_nonzeros()

auto get_jac_g_num_nonzeros

[Optional] Function that gets the number of nonzeros of the sparse Jacobian of the constraints.

Should return -1 for a dense Jacobian.

Required for second-order solvers only.

Definition at line 781 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_grad_gi()

void eval_grad_gi	(	crvec	x,
		index_t	i,
		rvec	grad_gi
	)		const

[Optional] Function that evaluates the gradient of one specific constraint, \( \nabla g_i(x) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	i	Which constraint \( 0 \le i \lt m \)
[out]	grad_gi	Gradient of the constraint \( \nabla g_i(x) \in \R^n \)

Required for second-order solvers only.

Definition at line 772 of file type-erased-problem.hpp.

◆ eval_hess_L_prod()

void eval_hess_L_prod	(	crvec	x,
		crvec	y,
		real_t	scale,
		crvec	v,
		rvec	Hv
	)		const

[Optional] Function that evaluates the Hessian of the Lagrangian multiplied by a vector, \( \nabla_{xx}^2L(x, y)\,v \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	y	Lagrange multipliers \( y \in \R^m \)
[in]	scale	Scale factor for the cost function.
[in]	v	Vector to multiply by \( v \in \R^n \)
[out]	Hv	Hessian-vector product \( \nabla_{xx}^2 L(x, y)\,v \in \R^{n} \)

Required for second-order solvers only.

Definition at line 785 of file type-erased-problem.hpp.

◆ eval_hess_L()

void eval_hess_L	(	crvec	x,
		crvec	y,
		real_t	scale,
		rindexvec	inner_idx,
		rindexvec	outer_ptr,
		rvec	H_values
	)		const

[Optional] Function that evaluates the Hessian of the Lagrangian as a sparse matrix, \( \nabla_{xx}^2L(x, y) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	y	Lagrange multipliers \( y \in \R^m \)
[in]	scale	Scale factor for the cost function.
[in,out]	inner_idx	Inner indices (row indices of nonzeros).
[in,out]	outer_ptr	Outer pointers (points to the first nonzero in each column).
[out]	H_values	Nonzero values of the Hessian \( \nabla_{xx}^2 L(x, y) \in \R^{n\times n} \). If `H_values` has size zero, this function should initialize `inner_idx` and `outer_ptr`. If `H_values` is nonempty, `inner_idx` and `outer_ptr` can be assumed to be initialized, and this function should evaluate `H_values`.

Required for second-order solvers only.

Definition at line 790 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_hess_L_num_nonzeros()

auto get_hess_L_num_nonzeros

[Optional] Function that gets the number of nonzeros of the sparse Hessian of the Lagrangian.

Should return -1 for a dense Hessian.

Required for second-order solvers only.

Definition at line 796 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_hess_ψ_prod()

void eval_hess_ψ_prod	(	crvec	x,
		crvec	y,
		crvec	Σ,
		real_t	scale,
		crvec	v,
		rvec	Hv
	)		const

[Optional] Function that evaluates the Hessian of the augmented Lagrangian multiplied by a vector, \( \nabla_{xx}^2L_\Sigma(x, y)\,v \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	y	Lagrange multipliers \( y \in \R^m \)
[in]	Σ	Penalty weights \( \Sigma \)
[in]	scale	Scale factor for the cost function.
[in]	v	Vector to multiply by \( v \in \R^n \)
[out]	Hv	Hessian-vector product \( \nabla_{xx}^2 L_\Sigma(x, y)\,v \in \R^{n} \)

Required for second-order solvers only.

Definition at line 800 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_hess_ψ()

void eval_hess_ψ	(	crvec	x,
		crvec	y,
		crvec	Σ,
		real_t	scale,
		rindexvec	inner_idx,
		rindexvec	outer_ptr,
		rvec	H_values
	)		const

[Optional] Function that evaluates the Hessian of the augmented Lagrangian, \( \nabla_{xx}^2L_\Sigma(x, y) \)

Parameters

[in]	x	Decision variable \( x \in \R^n \)
[in]	y	Lagrange multipliers \( y \in \R^m \)
[in]	Σ	Penalty weights \( \Sigma \)
[in]	scale	Scale factor for the cost function.
[in,out]	inner_idx	Inner indices (row indices of nonzeros).
[in,out]	outer_ptr	Outer pointers (points to the first nonzero in each column).
[out]	H_values	Nonzero values of the Hessian \( \nabla_{xx}^2 L_\Sigma(x, y) \in \R^{n\times n} \) If `H_values` has size zero, this function should initialize `inner_idx` and `outer_ptr`. If `H_values` is nonempty, `inner_idx` and `outer_ptr` can be assumed to be initialized, and this function should evaluate `H_values`.

Required for second-order solvers only.

Definition at line 805 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ get_hess_ψ_num_nonzeros()

auto get_hess_ψ_num_nonzeros

[Optional] Function that gets the number of nonzeros of the Hessian of the augmented Lagrangian.

Required for second-order solvers only.

Definition at line 811 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_f_grad_f()

auto eval_f_grad_f	(	crvec	x,
		rvec	grad_fx
	)		const

[Optional] Evaluate both \( f(x) \) and its gradient, \( \nabla f(x) \).

Default implementation:: ProblemVTable::default_eval_f_grad_f

Definition at line 815 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_f_g()

auto eval_f_g	(	crvec	x,
		rvec	g
	)		const

[Optional] Evaluate both \( f(x) \) and \( g(x) \).

Default implementation:: ProblemVTable::default_eval_f_g

Definition at line 819 of file type-erased-problem.hpp.

◆ eval_grad_f_grad_g_prod()

void eval_grad_f_grad_g_prod	(	crvec	x,
		crvec	y,
		rvec	grad_f,
		rvec	grad_gxy
	)		const

[Optional] Evaluate both \( \nabla f(x) \) and \( \nabla g(x)\,y \).

Default implementation:: ProblemVTable::default_eval_grad_f_grad_g_prod

Definition at line 823 of file type-erased-problem.hpp.

◆ eval_grad_L()

void eval_grad_L	(	crvec	x,
		crvec	y,
		rvec	grad_L,
		rvec	work_n
	)		const

[Optional] Evaluate the gradient of the Lagrangian \( \nabla_x L(x, y) = \nabla f(x) + \nabla g(x)\,y \)

Default implementation:: ProblemVTable::default_eval_grad_L

Definition at line 828 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_ψ()

auto eval_ψ	(	crvec	x,
		crvec	y,
		crvec	Σ,
		rvec	ŷ
	)		const

[Optional] Calculate both ψ(x) and the vector ŷ that can later be used to compute ∇ψ.

\[ \psi(x) = f(x) + \tfrac{1}{2} \text{dist}_\Sigma^2\left(g(x) + \Sigma^{-1}y,\;D\right) \]

\[ \hat y = \Sigma\, \left(g(x) + \Sigma^{-1}y - \Pi_D\left(g(x) + \Sigma^{-1}y\right)\right) \]

Default implementation:: ProblemVTable::default_eval_ψ

Parameters

[in]	x	Decision variable \( x \)
[in]	y	Lagrange multipliers \( y \)
[in]	Σ	Penalty weights \( \Sigma \)
[out]	ŷ	\( \hat y \)

Definition at line 833 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_grad_ψ()

void eval_grad_ψ	(	crvec	x,
		crvec	y,
		crvec	Σ,
		rvec	grad_ψ,
		rvec	work_n,
		rvec	work_m
	)		const

[Optional] Calculate the gradient ∇ψ(x).

\[ \nabla \psi(x) = \nabla f(x) + \nabla g(x)\,\hat y(x) \]

Default implementation:: ProblemVTable::default_eval_grad_ψ

Parameters

[in]	x	Decision variable \( x \)
[in]	y	Lagrange multipliers \( y \)
[in]	Σ	Penalty weights \( \Sigma \)
[out]	grad_ψ	\( \nabla \psi(x) \)
	work_n	Dimension \( n \)
	work_m	Dimension \( m \)

Definition at line 837 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ eval_ψ_grad_ψ()

auto eval_ψ_grad_ψ	(	crvec	x,
		crvec	y,
		crvec	Σ,
		rvec	grad_ψ,
		rvec	work_n,
		rvec	work_m
	)		const

[Optional] Calculate both ψ(x) and its gradient ∇ψ(x).

\[ \psi(x) = f(x) + \tfrac{1}{2} \text{dist}_\Sigma^2\left(g(x) + \Sigma^{-1}y,\;D\right) \]

\[ \nabla \psi(x) = \nabla f(x) + \nabla g(x)\,\hat y(x) \]

Default implementation:: ProblemVTable::default_eval_ψ_grad_ψ

Parameters

[in]	x	Decision variable \( x \)
[in]	y	Lagrange multipliers \( y \)
[in]	Σ	Penalty weights \( \Sigma \)
[out]	grad_ψ	\( \nabla \psi(x) \)
	work_n	Dimension \( n \)
	work_m	Dimension \( m \)

Definition at line 842 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ check()

void check

[Optional] Check that the problem formulation is well-defined, the dimensions match, etc.

Throws an exception if this is not the case.

Definition at line 859 of file type-erased-problem.hpp.

◆ provides_eval_inactive_indices_res_lna()

bool provides_eval_inactive_indices_res_lna ( ) const

inline

Returns true if the problem provides an implementation of eval_inactive_indices_res_lna.

Definition at line 591 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_jac_g()

bool provides_eval_jac_g ( ) const

inline

Returns true if the problem provides an implementation of eval_jac_g.

Definition at line 596 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_get_jac_g_num_nonzeros()

bool provides_get_jac_g_num_nonzeros ( ) const

inline

Returns true if the problem provides an implementation of get_jac_g_num_nonzeros.

Definition at line 601 of file type-erased-problem.hpp.

◆ provides_eval_grad_gi()

bool provides_eval_grad_gi ( ) const

inline

Returns true if the problem provides an implementation of eval_grad_gi.

Definition at line 606 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_hess_L_prod()

bool provides_eval_hess_L_prod ( ) const

inline

Returns true if the problem provides an implementation of eval_hess_L_prod.

Definition at line 611 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_hess_L()

bool provides_eval_hess_L ( ) const

inline

Returns true if the problem provides an implementation of eval_hess_L.

Definition at line 616 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_get_hess_L_num_nonzeros()

bool provides_get_hess_L_num_nonzeros ( ) const

inline

Returns true if the problem provides an implementation of get_hess_L_num_nonzeros.

Definition at line 621 of file type-erased-problem.hpp.

◆ provides_eval_hess_ψ_prod()

bool provides_eval_hess_ψ_prod ( ) const

inline

Returns true if the problem provides an implementation of eval_hess_ψ_prod.

Definition at line 626 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_hess_ψ()

bool provides_eval_hess_ψ ( ) const

inline

Returns true if the problem provides an implementation of eval_hess_ψ.

Definition at line 631 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_get_hess_ψ_num_nonzeros()

bool provides_get_hess_ψ_num_nonzeros ( ) const

inline

Returns true if the problem provides an implementation of get_hess_ψ_num_nonzeros.

Definition at line 636 of file type-erased-problem.hpp.

◆ provides_eval_f_grad_f()

bool provides_eval_f_grad_f ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_f_grad_f, false if it uses the default implementation.

Definition at line 641 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_f_g()

bool provides_eval_f_g ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_f_g, false if it uses the default implementation.

Definition at line 646 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_grad_f_grad_g_prod()

bool provides_eval_grad_f_grad_g_prod ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_grad_f_grad_g_prod, false if it uses the default implementation.

Definition at line 651 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_grad_L()

bool provides_eval_grad_L ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_grad_L, false if it uses the default implementation.

Definition at line 656 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_ψ()

bool provides_eval_ψ ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_ψ, false if it uses the default implementation.

Definition at line 661 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_grad_ψ()

bool provides_eval_grad_ψ ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_grad_ψ, false if it uses the default implementation.

Definition at line 664 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_eval_ψ_grad_ψ()

bool provides_eval_ψ_grad_ψ ( ) const

inline

Returns true if the problem provides a specialized implementation of eval_ψ_grad_ψ, false if it uses the default implementation.

Definition at line 669 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_get_box_C()

bool provides_get_box_C ( ) const

inline

Returns true if the problem provides an implementation of get_box_C.

Definition at line 674 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_get_box_D()

bool provides_get_box_D ( ) const

inline

Returns true if the problem provides an implementation of get_box_D.

Definition at line 679 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ provides_check()

bool provides_check ( ) const

inline

Returns true if the problem provides an implementation of check.

Definition at line 683 of file type-erased-problem.hpp.

Here is the caller graph for this function:

◆ calc_ŷ_dᵀŷ()

auto calc_ŷ_dᵀŷ	(	rvec	g_ŷ,
		crvec	y,
		crvec	Σ
	)		const

Given g(x), compute the intermediate results ŷ and dᵀŷ that can later be used to compute ψ(x) and ∇ψ(x).

Computes the result using the following algorithm:

\[ \begin{aligned} \zeta &= g(x) + \Sigma^{-1} y \\[] d &= \zeta - \Pi_D(\zeta) = \operatorname{eval\_proj\_diff\_g}(\zeta, \zeta) \\[] \hat y &= \Sigma d \\[] \end{aligned} \]

See also: Math

Parameters

[in,out]	g_ŷ	Input \( g(x) \), outputs \( \hat y \)
[in]	y	Lagrange multipliers \( y \)
[in]	Σ	Penalty weights \( \Sigma \)

Returns: The inner product \( d^\top \hat y \)

Definition at line 847 of file type-erased-problem.hpp.

◆ TypeErased() [1/10]

TypeErased ( )

defaultnoexcept

Default constructor.

◆ TypeErased() [2/10]

TypeErased	(	std::allocator_arg_t	,
		const Alloc &	alloc
	)

inline

Default constructor (allocator aware).

Definition at line 225 of file type-erasure.hpp.

◆ TypeErased() [3/10]

TypeErased ( const TypeErased & other )

inline

Copy constructor.

Definition at line 227 of file type-erasure.hpp.

◆ TypeErased() [4/10]

TypeErased	(	const TypeErased &	other,
		allocator_type	alloc
	)

inline

Copy constructor (allocator aware).

Definition at line 233 of file type-erasure.hpp.

◆ TypeErased() [5/10]

TypeErased ( TypeErased && other )

inlinenoexcept

Move constructor.

Definition at line 249 of file type-erasure.hpp.

◆ TypeErased() [6/10]

TypeErased	(	TypeErased &&	other,
		const allocator_type &	alloc
	)

inlinenoexcept

Move constructor (allocator aware).

Definition at line 267 of file type-erasure.hpp.

◆ TypeErased() [7/10]

TypeErased	(	std::allocator_arg_t	,
		const Alloc &	alloc,
		T &&	d
	)

inlineexplicit

Main constructor that type-erases the given argument.

Definition at line 352 of file type-erasure.hpp.

◆ TypeErased() [8/10]

TypeErased	(	std::allocator_arg_t	,
		const Alloc &	alloc,
		te_in_place_t< T >	,
		Args &&...	args
	)

inlineexplicit

Main constructor that type-erases the object constructed from the given argument.

Definition at line 359 of file type-erasure.hpp.

◆ TypeErased() [9/10]

TypeErased ( T && d )

inlineexplicit

Requirement prevents this constructor from taking precedence over the copy and move constructors.

Definition at line 369 of file type-erasure.hpp.

◆ TypeErased() [10/10]

TypeErased	(	te_in_place_t< T >	,
		Args &&...	args
	)

inlineexplicit

Main constructor that type-erases the object constructed from the given argument.

Definition at line 375 of file type-erasure.hpp.

◆ call() [1/8]

decltype(auto) call	(	Ret()(const void , FArgs...)	f,
		Args &&...	args
	)		const

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 514 of file type-erasure.hpp.

◆ call() [2/8]

decltype(auto) call	(	Ret()(void , FArgs...)	f,
		Args &&...	args
	)

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 526 of file type-erasure.hpp.

◆ call() [3/8]

decltype(auto) call ( Ret(*)(const void *) f ) const

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 538 of file type-erasure.hpp.

◆ call() [4/8]

decltype(auto) call ( Ret(*)(void *) f )

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 545 of file type-erasure.hpp.

◆ call() [5/8]

decltype(auto) call ( Ret(*)(const void *, const VTable &) f ) const

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 552 of file type-erasure.hpp.

◆ call() [6/8]

decltype(auto) call ( Ret(*)(void *, const VTable &) f )

inlineprotected

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 560 of file type-erasure.hpp.

◆ make() [2/2]

static Ret make	(	std::allocator_arg_t	tag,
		const Alloc &	alloc,
		Args &&...	args
	)

inlinestaticinherited

Construct a type-erased wrapper of type Ret for an object of type T, initialized in-place with the given arguments.

Definition at line 383 of file type-erasure.hpp.

◆ operator bool()

operator bool ( ) const

inlineexplicitnoexceptinherited

Check if this wrapper wraps an object.

False for default-constructed objects.

Definition at line 400 of file type-erasure.hpp.

◆ get_allocator()

allocator_type get_allocator ( ) const

inlinenoexceptinherited

Get a copy of the allocator.

Definition at line 403 of file type-erasure.hpp.

◆ type()

const std::type_info & type ( ) const

inlinenoexceptinherited

Query the contained type.

Definition at line 406 of file type-erasure.hpp.

Here is the caller graph for this function:

◆ as() [1/3]

T & as ( ) &

inlineinherited

Convert the type-erased object to the given type.

The type is checked in debug builds only, use with caution.

Definition at line 413 of file type-erasure.hpp.

Here is the call graph for this function:

◆ as() [2/3]

const T & as ( ) const &

inlineinherited

Convert the type-erased object to the given type.

The type is checked in debug builds only, use with caution.

Definition at line 420 of file type-erasure.hpp.

Here is the call graph for this function:

◆ as() [3/3]

const T && as ( ) &&

inlineinherited

Convert the type-erased object to the given type.

The type is checked in debug builds only, use with caution.

Definition at line 427 of file type-erasure.hpp.

Here is the call graph for this function:

◆ allocate()

Deallocator allocate ( size_t size )

inlineprivateinherited

Ensure that storage is available, either by using the small buffer if it is large enough, or by calling the allocator.

Returns a RAII wrapper that deallocates the storage unless released.

Definition at line 450 of file type-erasure.hpp.

Here is the caller graph for this function:

◆ deallocate()

void deallocate ( )

inlineprivateinherited

Deallocate the memory without invoking the destructor.

Definition at line 460 of file type-erasure.hpp.

Here is the caller graph for this function:

◆ cleanup()

void cleanup ( )

inlineprivateinherited

Destroy the type-erased object (if not empty), and deallocate the memory if necessary.

Definition at line 470 of file type-erasure.hpp.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ do_copy_assign()

void do_copy_assign ( const TypeErased< VTable, Allocator, SmallBufferSize > & other )

inlineprivateinherited

Definition at line 478 of file type-erasure.hpp.

Here is the call graph for this function:

◆ construct_inplace()

void construct_inplace ( Args &&... args )

inlineprotectedinherited

Ensure storage and construct the type-erased object of type T in-place.

Definition at line 498 of file type-erasure.hpp.

Here is the call graph for this function:

◆ call() [7/8]

decltype(auto) call ( Ret(*)(const void *, const VTable &) f ) const

inlineprotectedinherited

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 552 of file type-erasure.hpp.

◆ call() [8/8]

decltype(auto) call ( Ret(*)(void *, const VTable &) f )

inlineprotectedinherited

Call the vtable function f with the given arguments args, implicitly passing the self pointer and vtable reference if necessary.

Definition at line 560 of file type-erasure.hpp.

Member Data Documentation

◆ self

void* self

protected

Pointer to the stored object.

Definition at line 215 of file type-erasure.hpp.

◆ vtable

VTable vtable

protected

Definition at line 218 of file type-erasure.hpp.

◆ small_buffer_size

constexpr size_t small_buffer_size = SmallBufferSize

staticconstexprinherited

Definition at line 196 of file type-erasure.hpp.

◆ small_buffer

buffer_type small_buffer

privateinherited

Definition at line 202 of file type-erasure.hpp.

◆ allocator

allocator_type allocator

privateinherited

Definition at line 203 of file type-erasure.hpp.

◆ no_child_of_ours

constexpr auto no_child_of_ours

staticconstexprprivateinherited

Initial value:

=

!std::is_base_of_v<TypeErased, std::remove_cvref_t<T>>

True if T is not a child class of TypeErased.

Definition at line 208 of file type-erasure.hpp.

◆ invalid_size

constexpr size_t invalid_size

staticconstexprprotectedinherited

Initial value:

=

static_cast<size_t>(0xDEADBEEFDEADBEEF)

Definition at line 212 of file type-erasure.hpp.

◆ size

size_t size = invalid_size

protectedinherited

Size required to store the object.

Definition at line 217 of file type-erasure.hpp.

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

src/alpaqa/include/alpaqa/problem/type-erased-problem.hpp

Detailed Description

Problem dimensions

Required cost and constraint functions

Projections onto constraint sets and proximal mappings

Constraint sets

Functions for second-order solvers

Combined evaluations

Augmented Lagrangian

Checks

Querying specialized implementations

Helpers

Public Types

Public Member Functions

Static Public Member Functions

Static Public Attributes

Protected Member Functions

Protected Attributes

Static Protected Attributes

Private Types

Private Member Functions

Private Attributes

Static Private Attributes

Member Typedef Documentation

◆ Box

◆ VTable

◆ allocator_type

◆ TypeErased

◆ allocator_traits

◆ buffer_type

Member Function Documentation

◆ make() [1/2]

◆ get_n()

◆ get_m()

◆ eval_f()

◆ eval_grad_f()

◆ eval_g()

◆ eval_grad_g_prod()

◆ eval_proj_diff_g()

◆ eval_proj_multipliers()

◆ eval_prox_grad_step()

◆ eval_inactive_indices_res_lna()

◆ get_box_C()

◆ get_box_D()

◆ eval_jac_g()

◆ get_jac_g_num_nonzeros()

◆ eval_grad_gi()

◆ eval_hess_L_prod()

◆ eval_hess_L()

◆ get_hess_L_num_nonzeros()

◆ eval_hess_ψ_prod()

◆ eval_hess_ψ()

◆ get_hess_ψ_num_nonzeros()

◆ eval_f_grad_f()

◆ eval_f_g()

◆ eval_grad_f_grad_g_prod()

◆ eval_grad_L()

◆ eval_ψ()

◆ eval_grad_ψ()

◆ eval_ψ_grad_ψ()

◆ check()

◆ provides_eval_inactive_indices_res_lna()

◆ provides_eval_jac_g()

◆ provides_get_jac_g_num_nonzeros()

◆ provides_eval_grad_gi()

◆ provides_eval_hess_L_prod()

◆ provides_eval_hess_L()

◆ provides_get_hess_L_num_nonzeros()

◆ provides_eval_hess_ψ_prod()

◆ provides_eval_hess_ψ()

◆ provides_get_hess_ψ_num_nonzeros()

◆ provides_eval_f_grad_f()

◆ provides_eval_f_g()

◆ provides_eval_grad_f_grad_g_prod()

◆ provides_eval_grad_L()

◆ provides_eval_ψ()

◆ provides_eval_grad_ψ()

◆ provides_eval_ψ_grad_ψ()

◆ provides_get_box_C()

◆ provides_get_box_D()

◆ provides_check()