This class store the adjacent processors and the adjacent sub_domains. More...

Detailed Description

template<unsigned int dim, typename T>
class nn_prcs< dim, T >

This class store the adjacent processors and the adjacent sub_domains.

Template Parameters

dim	is the dimensionality of the physical domain we are going to decompose.
T	type of the space we decompose, Real, Integer, Complex ...

See Also: CartDecomposition

Definition at line 23 of file nn_processor.hpp.

#include <nn_processor.hpp>

Inheritance diagram for nn_prcs< dim, T >:

Public Member Functions
	nn_prcs (Vcluster &v_cl)
	Constructor require Vcluster.

	nn_prcs (const nn_prcs< dim, T > &ilg)
	Constructor from another nn_prcs.

	nn_prcs (nn_prcs< dim, T > &&ilg)
	Constructor from temporal ie_loc_ghost.

nn_prcs< dim, T > &	operator= (const nn_prcs< dim, T > &nnp)
	Copy the object. More...

nn_prcs< dim, T > &	operator= (nn_prcs< dim, T > &&nnp)
	Copy the object. More...

void	create (const openfpm::vector< openfpm::vector< long unsigned int > > &box_nn_processor, const openfpm::vector< SpaceBox< dim, T >> &sub_domains)
	Create the list of adjacent processors and the list of adjacent sub-domains. More...

size_t	getNNProcessors () const
	Get the number of Near processors. More...

size_t	IDtoProc (size_t id) const
	Return the processor id of the near processor list at place id. More...

const openfpm::vector< size_t > &	getNearSubdomainsRealId (size_t p_id) const
	Get the real-id of the sub-domains of a near processor. More...

const openfpm::vector< ::Box < dim, T > > &	getNearSubdomains (size_t p_id) const
	Get the sub-domains of a near processor. More...

size_t	getNRealSubdomains (size_t p_id) const
	Get the number of real sub-domains of a near processor. More...

const openfpm::vector< comb < dim > > &	getNearSubdomainsPos (size_t p_id) const
	Get the sub-domains sector position of a near processor. More...

size_t	getNearProcessor (size_t p_id) const
	Get the near processor id. More...

const openfpm::vector< size_t > &	getSentSubdomains (size_t p_id) const
	For each near processor it give a vector with the id of the local sub-domain sent to that processor. More...

size_t	ProctoID (size_t p) const
	Convert the processor rank to the id in the list. More...

bool	write (std::string output) const
	Write the decomposition as VTK file. More...

void	applyBC (const Box< dim, T > &domain, const Ghost< dim, T > &ghost, const size_t(&bc)[dim])
	Apply boundary conditions. More...

bool	is_equal (nn_prcs< dim, T > &np)
	Check if the nn_prcs contain the same information. More...

void	reset ()
	Reset the nn_prcs structure. More...

std::unordered_map< size_t, N_box< dim, T > > &	get_nn_processor_subdomains ()
	Used for testing porpose do not use.

openfpm::vector< size_t > &	get_nn_processors ()
	Used for testing porpose do not use.

Static Public Member Functions
static bool	check_valid (comb< dim > cmb, const size_t(&bc)[dim])

Private Member Functions
void	consistent_shift (Box< dim, T > &box, const Box< dim, T > &domain, const Point< dim, T > &shift)
	It shift a box but it does consistently. More...

void	add_nn_subdomain (size_t i, size_t r_sub, const Box< dim, T > &bx, const comb< dim > &c)
	add sub-domains to processor for a near processor i More...

void	add_box_periodic (const Box< dim, T > &domain, const Ghost< dim, T > &ghost, const size_t(&bc)[dim])
	In case of periodic boundary conditions we replicate the sub-domains at the border. More...

void	flush ()
	Flush the temporal added sub-domain to the processor sub-domain. More...

Static Private Member Functions
static void *	message_alloc (size_t msg_i, size_t total_msg, size_t total_p, size_t i, size_t ri, void *ptr)
	Message allocation. More...

Private Attributes
Vcluster &	v_cl
	Virtual cluster.

openfpm::vector< size_t >	nn_processors
	List of adjacent processors.

std::unordered_map< size_t, N_box< dim, T > >	nn_processor_subdomains
	for each near processor store the sub-domains of the near processors

std::unordered_map< size_t, N_box< dim, T > >	nn_processor_subdomains_tmp
	when we add new boxes, are added here

openfpm::vector < openfpm::vector< size_t > >	proc_adj_box

openfpm::vector < openfpm::vector< ::SpaceBox < dim, T > > >	boxes
	contain the set of sub-domains sent to the other processors

size_t	recv_cnt
	Receive counter.

bool	aBC
	applyBC function is suppose to be called only one time

Member Function Documentation

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::add_box_periodic	(	const Box< dim, T > &	domain,
		const Ghost< dim, T > &	ghost,
		const size_t(&)	bc[dim]
	)

inlineprivate

In case of periodic boundary conditions we replicate the sub-domains at the border.

Parameters

domain	Domain
ghost	ghost part
bc	boundary boundary conditions

Definition at line 173 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::add_nn_subdomain	(	size_t	i,
		size_t	r_sub,
		const Box< dim, T > &	bx,
		const comb< dim > &	c
	)

inlineprivate

add sub-domains to processor for a near processor i

Parameters

i	near processor
r_sub	real sub-domain id
bx	Box to add
c	from which sector the sub-domain come from

Definition at line 158 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::applyBC	(	const Box< dim, T > &	domain,
		const Ghost< dim, T > &	ghost,
		const size_t(&)	bc[dim]
	)

inline

Apply boundary conditions.

Parameters

domain	The simulation domain
ghost	ghost part
bc	Boundary conditions

Definition at line 628 of file nn_processor.hpp.

template<unsigned int dim, typename T>

static bool nn_prcs< dim, T >::check_valid	(	comb< dim >	cmb,
		const size_t(&)	bc[dim]
	)

inlinestatic

Check that the combination is valid

Is a function that is used in otder to understand if a sub-domain must be mirrored because of boundary conditions

Parameters

cmb	combination
bc	boundary conditions

Returns: true if the combination is valid

Definition at line 303 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::consistent_shift	(	Box< dim, T > &	box,
		const Box< dim, T > &	domain,
		const Point< dim, T > &	shift
	)

inlineprivate

It shift a box but it does consistently.

In calculating internal and external ghost boxes, domains are shifted by periodicity. In particular, consider a box touching with the left bolder the left border of the domain

 before shift                              after shift

+--------------------------—+ +---------------------------—+ | | | |

domain		domain

+------—+		+------—+


box				box

+------—+		+------—+

+--------------------------—+ +---------------------------—+

   shifting the box on the right by the size of the domain, we expect to have a box touching with
   the left side the right side of the domain. Because of rounding off problem this is not possible
   with a simple shift. This function ensure consistency like ensuring the previous condition, with
   the assumption that the shift is +/- the domain size

   \param box to shift
   \param domain
   \param shift

Definition at line 93 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::create	(	const openfpm::vector< openfpm::vector< long unsigned int > > &	box_nn_processor,
		const openfpm::vector< SpaceBox< dim, T >> &	sub_domains
	)

inline

Create the list of adjacent processors and the list of adjacent sub-domains.

Parameters

box_nn_processor	list of adjacent processors for each sub-domain
sub_domains	list of local sub-domains

Definition at line 357 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::flush ( )

inlineprivate

Flush the temporal added sub-domain to the processor sub-domain.

Definition at line 251 of file nn_processor.hpp.

template<unsigned int dim, typename T>

size_t nn_prcs< dim, T >::getNearProcessor ( size_t p_id ) const

inline

Get the near processor id.

Parameters

p_id	adjacent processor rank

Returns: the processor rank

Definition at line 546 of file nn_processor.hpp.

template<unsigned int dim, typename T>

const openfpm::vector< ::Box<dim,T> >& nn_prcs< dim, T >::getNearSubdomains ( size_t p_id ) const

inline

Get the sub-domains of a near processor.

Parameters

p_id	near processor rank

Returns: the sub-domains

Definition at line 485 of file nn_processor.hpp.

template<unsigned int dim, typename T>

const openfpm::vector< comb<dim> >& nn_prcs< dim, T >::getNearSubdomainsPos ( size_t p_id ) const

inline

Get the sub-domains sector position of a near processor.

Parameters

p_id	near processor rank

Returns: the sub-domains positions

Definition at line 527 of file nn_processor.hpp.

template<unsigned int dim, typename T>

const openfpm::vector< size_t >& nn_prcs< dim, T >::getNearSubdomainsRealId ( size_t p_id ) const

inline

Get the real-id of the sub-domains of a near processor.

Parameters

p_id	near processor rank

Returns: the sub-domains real id

Definition at line 465 of file nn_processor.hpp.

template<unsigned int dim, typename T>

size_t nn_prcs< dim, T >::getNNProcessors ( ) const

inline

Get the number of Near processors.

Returns: the number of near processors

Definition at line 441 of file nn_processor.hpp.

template<unsigned int dim, typename T>

size_t nn_prcs< dim, T >::getNRealSubdomains ( size_t p_id ) const

inline

Get the number of real sub-domains of a near processor.

Note: the real sub-domain are the subdomain in the central sector, or any sub-domain that has not been create because of boundary conditions

Parameters

p_id	near processor rank

Returns: the number of real sub-domains

Definition at line 507 of file nn_processor.hpp.

template<unsigned int dim, typename T>

const openfpm::vector<size_t>& nn_prcs< dim, T >::getSentSubdomains ( size_t p_id ) const

inline

For each near processor it give a vector with the id of the local sub-domain sent to that processor.

Parameters

p_id	adjacent processor (id from 0 to getNNProcessors())

Returns: a vector of sub-domains id

Definition at line 567 of file nn_processor.hpp.

template<unsigned int dim, typename T>

size_t nn_prcs< dim, T >::IDtoProc ( size_t id ) const

inline

Return the processor id of the near processor list at place id.

Parameters

id

Returns: return the processor rank

Definition at line 453 of file nn_processor.hpp.

template<unsigned int dim, typename T>

bool nn_prcs< dim, T >::is_equal ( nn_prcs< dim, T > & np )

inline

Check if the nn_prcs contain the same information.

Parameters

np	Element to check

Returns: true if they are equal

Definition at line 648 of file nn_processor.hpp.

template<unsigned int dim, typename T>

static void* nn_prcs< dim, T >::message_alloc	(	size_t	msg_i,
		size_t	total_msg,
		size_t	total_p,
		size_t	i,
		size_t	ri,
		void *	ptr
	)

inlinestaticprivate

Message allocation.

Parameters

msg_i	message size required to receive from i
total_msg	total message size to receive from all the processors
total_p	the total number of processor want to communicate with you
i	processor id from which we receive
ri	request id (it is an id that goes from 0 to total_p, and is unique every time message_alloc is called)
ptr	a pointer to the vector_dist structure

Returns: the pointer where to store the message

Definition at line 139 of file nn_processor.hpp.

template<unsigned int dim, typename T>

nn_prcs<dim,T>& nn_prcs< dim, T >::operator= ( const nn_prcs< dim, T > & nnp )

inline

Copy the object.

Parameters

nnp	object to copy

Returns: itself

Definition at line 324 of file nn_processor.hpp.

template<unsigned int dim, typename T>

nn_prcs<dim,T>& nn_prcs< dim, T >::operator= ( nn_prcs< dim, T > && nnp )

inline

Copy the object.

Parameters

nnp	object to copy

Returns: itself

Definition at line 341 of file nn_processor.hpp.

template<unsigned int dim, typename T>

size_t nn_prcs< dim, T >::ProctoID ( size_t p ) const

inline

Convert the processor rank to the id in the list.

Parameters

p	processor rank

Returns: the id

Definition at line 579 of file nn_processor.hpp.

template<unsigned int dim, typename T>

void nn_prcs< dim, T >::reset ( )

inline

Reset the nn_prcs structure.

Definition at line 669 of file nn_processor.hpp.

template<unsigned int dim, typename T>

bool nn_prcs< dim, T >::write ( std::string output ) const

inline

Write the decomposition as VTK file.

The function generate several files

1) subdomains_adjacent_X.vtk sub-domains adjacent to the local processor (X)

where X is the local processor rank

Parameters

output directory where to write the files

Returns: true if the write procedure succeed

subdomains_adjacent_X.vtk sub-domains adjacent to the local processor (X)

Definition at line 605 of file nn_processor.hpp.

Field Documentation

template<unsigned int dim, typename T>

openfpm::vector<openfpm::vector<size_t> > nn_prcs< dim, T >::proc_adj_box

private

contain the same information as the member boxes with the difference that instead of the Box itself, it contain the sub-domain id in the list of the local sub-domains

Definition at line 40 of file nn_processor.hpp.

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

src/Decomposition/nn_processor.hpp

Detailed Description

template<unsigned int dim, typename T> class nn_prcs< dim, T >

Public Member Functions

Static Public Member Functions

Private Member Functions

Static Private Member Functions

Private Attributes

Member Function Documentation

Field Documentation

template<unsigned int dim, typename T>
class nn_prcs< dim, T >