QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace > Class Template Reference#

DiFfRG: DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace > Class Template Reference

DiFfRG

Discretization Framework for functional Renormalization Group flows

This class performs numerical integration over a d-dimensional hypercube using quadrature rules. More...

#include <quadrature_integrator.hh>

Inheritance diagram for DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >:

Public Types
using	ctype = typename get_type::ctype<NT>
	Numerical type to be used for integration tasks e.g. the argument or possible jacobians.

using	execution_space = ExecutionSpace
	Execution space to be used for the integration, e.g. GPU_exec, TBB_exec.

Public Member Functions
	QuadratureIntegrator (QuadratureProvider &quadrature_provider, const std::array< size_t, dim > &_grid_size, const std::array< ctype, dim > &grid_min, const std::array< ctype, dim > &grid_max, const std::array< QuadratureType, dim > &quadrature_type)

void	set_grid_extents (const std::array< ctype, dim > &grid_min, const std::array< ctype, dim > &grid_max)

template<typename... T> requires is_valid_kernel<NT, KERNEL, ctype, dim, T...>
void	get (NT &dest, const T &...t) const

template<typename OT , typename... T> requires (!std::is_same_v<OT, NT> && is_valid_kernel<NT, KERNEL, ctype, dim, T...>)
void	get (OT &dest, const T &...t) const

template<typename OT , typename... T> requires (!std::is_same_v<OT, NT> && is_valid_kernel<NT, KERNEL, ctype, dim, T...>)
void	get (ExecutionSpace &space, OT &dest, const T &...t) const

template<typename view_type , typename Coordinates , typename... Args>
void	map (ExecutionSpace &space, const view_type integral_view, const Coordinates &coordinates, const Args &...args)

size_t	quadrature_volume () const
	Points evaluated per external grid point. Half of the scheduler's cost score.

template<typename Coordinates , typename... Args>
auto	map (NT *dest, const Coordinates &coordinates, const Args &...args)

template<typename Coordinates , typename... Args>
auto	map_dist (NT *dest, const Coordinates &coordinates, const Args &...args)

Public Member Functions inherited from DiFfRG::AbstractIntegrator
	AbstractIntegrator ()

KOKKOS_FORCEINLINE_FUNCTION size_t	integrator_id () const
	Stable, rank-independent identity of this integrator.

Protected Attributes
device::array< size_t, dim >	grid_size

ExecutionSpace	space

QuadratureProvider &	quadrature_provider

device::array< device::array< ctype, dim >, 2 >	grid_extents

device::array< ctype, dim >	grid_start

device::array< ctype, dim >	grid_scale

device::array< Kokkos::View< const ctype *, typename ExecutionSpace::memory_space >, dim >	nodes

device::array< Kokkos::View< const ctype *, typename ExecutionSpace::memory_space >, dim >	weights

KokkosNDView< 1+dim, NT, ExecutionSpace >	m_cache

device::array< size_t, 1+dim >	m_cache_extents {}

Kokkos::View< ctype *, typename ExecutionSpace::memory_space >	m_positions

std::string	m_positions_key

Kokkos::View< NT *, ExecutionSpace >	m_dest_device

size_t	m_dest_device_size = 0

Kokkos::View< NT *, PinnedHost_memory >	m_dest_pinned

size_t	m_dest_pinned_size = 0

Kokkos::View< NT, typename ExecutionSpace::memory_space >	m_result_view

Kokkos::View< NT, typenameExecutionSpace::memory_space >::host_mirror_type	m_result_host

bool	m_result_views_initialized = false

Protected Attributes inherited from DiFfRG::AbstractIntegrator
size_t	m_integrator_id

Private Member Functions
template<typename Coordinates , typename... Args>
void	run_or_queue_host (NT *dest, const Coordinates &coordinates, const Args &...args)
	Run a host-backend map now, or queue it for flush time if a deferral scope is open.

template<typename Coordinates , typename... Args>
void	run_host (NT *dest, const Coordinates &coordinates, const Args &...args)

Detailed Description

template<int dim, typename NT, typename KERNEL, typename ExecutionSpace>
requires (dim > 0)
class DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >

This class performs numerical integration over a d-dimensional hypercube using quadrature rules.

Template Parameters

dim	The dimension of the hypercube, which can be between 1 and 5.
NT	numerical type of the result
KERNEL	kernel to be integrated, which must provide the static methods `kernel` and `constant`
ExecutionSpace	can be any execution space, e.g. GPU_exec, TBB_exec.

Member Typedef Documentation

◆ ctype

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

using DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::ctype = typename get_type::ctype<NT>

Numerical type to be used for integration tasks e.g. the argument or possible jacobians.

◆ execution_space

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

using DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::execution_space = ExecutionSpace

Execution space to be used for the integration, e.g. GPU_exec, TBB_exec.

Constructor & Destructor Documentation

◆ QuadratureIntegrator()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::QuadratureIntegrator	(	QuadratureProvider &	quadrature_provider,
		const std::array< size_t, dim > &	_grid_size,
		const std::array< ctype, dim > &	grid_min,
		const std::array< ctype, dim > &	grid_max,
		const std::array< QuadratureType, dim > &	quadrature_type )

inline

Member Function Documentation

◆ get() [1/3]

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename OT , typename... T>
requires (!std::is_same_v<OT, NT> && is_valid_kernel<NT, KERNEL, ctype, dim, T...>)

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::get	(	ExecutionSpace &	space,
		OT &	dest,
		const T &...	t ) const

inline

◆ get() [2/3]

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename... T>
requires is_valid_kernel<NT, KERNEL, ctype, dim, T...>

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::get	(	NT &	dest,
		const T &...	t ) const

inline

◆ get() [3/3]

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename OT , typename... T>
requires (!std::is_same_v<OT, NT> && is_valid_kernel<NT, KERNEL, ctype, dim, T...>)

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::get	(	OT &	dest,
		const T &...	t ) const

inline

◆ map() [1/2]

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename view_type , typename Coordinates , typename... Args>

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::map	(	ExecutionSpace &	space,
		const view_type	integral_view,
		const Coordinates &	coordinates,
		const Args &...	args )

inline

◆ map() [2/2]

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename Coordinates , typename... Args>

auto DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::map	(	NT *	dest,
		const Coordinates &	coordinates,
		const Args &...	args )

inline

◆ map_dist()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename Coordinates , typename... Args>

auto DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::map_dist	(	NT *	dest,
		const Coordinates &	coordinates,
		const Args &...	args )

inline

◆ quadrature_volume()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

size_t DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::quadrature_volume ( ) const

inline

Points evaluated per external grid point. Half of the scheduler's cost score.

◆ run_host()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename Coordinates , typename... Args>

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::run_host	(	NT *	dest,
		const Coordinates &	coordinates,
		const Args &...	args )

inlineprivate

The host-backend body of map_dist(). Queued jobs run one after another, so the shared m_dest_device scratch is written and drained before the next job touches it.

◆ run_or_queue_host()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

template<typename Coordinates , typename... Args>

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::run_or_queue_host	(	NT *	dest,
		const Coordinates &	coordinates,
		const Args &...	args )

inlineprivate

Run a host-backend map now, or queue it for flush time if a deferral scope is open.

A host kernel is synchronous, so running it inline blocks the host thread and delays the launch of any device map issued after it. Queuing makes the call order inside a DeferredMaps scope irrelevant: flush() runs the queue before it fences, i.e. while the device work already launched is still running. See MapCompletion::record_work.

Compiled out entirely in a CUDA-less build, where there is no device to overlap with.

◆ set_grid_extents()

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

void DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::set_grid_extents	(	const std::array< ctype, dim > &	grid_min,
		const std::array< ctype, dim > &	grid_max )

inline

Member Data Documentation

◆ grid_extents

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<device::array<ctype, dim>, 2> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::grid_extents

protected

◆ grid_scale

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<ctype, dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::grid_scale

protected

◆ grid_size

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<size_t, dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::grid_size

protected

◆ grid_start

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<ctype, dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::grid_start

protected

◆ m_cache

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

KokkosNDView<1 + dim, NT, ExecutionSpace> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_cache

mutableprotected

◆ m_cache_extents

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<size_t, 1 + dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_cache_extents {}

mutableprotected

◆ m_dest_device

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

Kokkos::View<NT *, ExecutionSpace> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_dest_device

mutableprotected

◆ m_dest_device_size

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

size_t DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_dest_device_size = 0

mutableprotected

◆ m_dest_pinned

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

Kokkos::View<NT *, PinnedHost_memory> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_dest_pinned

mutableprotected

◆ m_dest_pinned_size

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

size_t DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_dest_pinned_size = 0

mutableprotected

◆ m_positions

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

Kokkos::View<ctype *, typename ExecutionSpace::memory_space> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_positions

mutableprotected

◆ m_positions_key

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

std::string DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_positions_key

mutableprotected

◆ m_result_host

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

Kokkos::View<NT,typenameExecutionSpace::memory_space>::host_mirror_type DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_result_host

mutableprotected

◆ m_result_view

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

Kokkos::View<NT, typename ExecutionSpace::memory_space> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_result_view

mutableprotected

◆ m_result_views_initialized

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

bool DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::m_result_views_initialized = false

mutableprotected

◆ nodes

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<Kokkos::View<const ctype *, typename ExecutionSpace::memory_space>, dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::nodes

protected

◆ quadrature_provider

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

QuadratureProvider& DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::quadrature_provider

protected

◆ space

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

ExecutionSpace DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::space

protected

◆ weights

template<int dim, typename NT , typename KERNEL , typename ExecutionSpace >

device::array<Kokkos::View<const ctype *, typename ExecutionSpace::memory_space>, dim> DiFfRG::QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace >::weights

protected

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

/home/runner/work/DiFfRG_current/DiFfRG_current/DiFfRG/include/DiFfRG/physics/integration/quadrature_integrator.hh

QuadratureIntegrator&lt; dim, NT, KERNEL, ExecutionSpace &gt; Class Template Reference#

Public Types

Public Member Functions

Protected Attributes

Private Member Functions

Detailed Description

Member Typedef Documentation

◆ ctype

◆ execution_space

Constructor & Destructor Documentation

◆ QuadratureIntegrator()

Member Function Documentation

◆ get() [1/3]

◆ get() [2/3]

◆ get() [3/3]

◆ map() [1/2]

◆ map() [2/2]

◆ map_dist()

◆ quadrature_volume()

◆ run_host()

◆ run_or_queue_host()

◆ set_grid_extents()

Member Data Documentation

◆ grid_extents

◆ grid_scale

◆ grid_size

◆ grid_start

◆ m_cache

◆ m_cache_extents

◆ m_dest_device

◆ m_dest_device_size

◆ m_dest_pinned

◆ m_dest_pinned_size

◆ m_positions

◆ m_positions_key

◆ m_result_host

◆ m_result_view

◆ m_result_views_initialized

◆ nodes

◆ quadrature_provider

◆ space

◆ weights

QuadratureIntegrator< dim, NT, KERNEL, ExecutionSpace > Class Template Reference#