NNSR Class Reference

Neighbor Node Sampling and Repeat algorithm. More...

#include <algo/NNSR.h>

Public Member Functions
ytNetwork *	NNSR_exec (Args *args)
	Performs the NNSR algorithm.
void	NNSR_setArgs (Args *args, NNSRWork *work)
	Sets the some params of Args to the NNSRWork instance.
int	NNSR_RandomWalk (int v, ytPCGraph *g, ytDoubleArray *paFreqs, ytDoubleArray *chFreqs, int size, int num, int *set, int setSize, int limit, int resetPosition, ytIntArray *W, ytDoubleArray *P, ytRNG *rng)
	Samples nodes by random walk.
int	NNSR_otoSampling (int i, int p, int MPI_size, int MPI_rank, int size, int num, int *set, int setSize, const double *oto, ytRNG *rng)
	Performs one-to-one sampling.
int	NNSR_nodeSampling (NNSRWork *work, int v, int *nodeSet, int nodeSetSize)
	Sampling nodes by random walk and one-to-one scores.
void	NNSR_collectData (Args *args, NNSRWork *work, int *nodeSet, int nodeSetSize)
	Collects the data of the selected nodes.
void	NNSR_shareNetwork (NNSRWork *work, double threshold)
	Shares a global network by collecting edges among processors.

Detailed Description

Neighbor Node Sampling and Repeat algorithm.

See Neighbor Node Sampling and Repeat Algorithm for the usage of the algorithm.

Member Function Documentation

NNSR_collectData()

void NNSR_collectData	(	Args *	args,
		NNSRWork *	work,
		int *	nodeSet,
		int	nodeSetSize
	)

Collects the data of the selected nodes.

This copies the samples of variables in nodeSet stored in NNSRWork::fullData, to Args::data given as work.

Parameters

args	Args instance.
work	NNSRWork instance.
nodeSet	set of node IDs to collect.
nodeSetSize	size of nodeSet.

NNSR_exec()

ytNetwork * NNSR_exec

(

Args *

args

)

Performs the NNSR algorithm.

This performs the Neighbor Node Sampling and Repeat (NNSR) algorithm, that searches for the Bayesian network structure by the MPI-based parallel algorithm. Thus, this requires MPI.

Only the root rank returns the final result of the network estimation.

NNSR_otoSampling()

int NNSR_otoSampling	(	int	i,
		int	p,
		int	MPI_size,
		int	MPI_rank,
		int	size,
		int	num,
		int *	set,
		int	setSize,
		const double *	oto,
		ytRNG *	rng
	)

Performs one-to-one sampling.

Parameters

	i	index of the target node in this process such that i=0, 1, ..., args->subNodes - 1.
	MPI_size	size of MPI processes.
	size	maximum size of set (size of the array).
	num	Number of nodes to sample.
[in,out]	set
	setSize	current size of set.

NNSR_RandomWalk()

int NNSR_RandomWalk	(	int	v,
		ytPCGraph *	g,
		ytDoubleArray *	paFreqs,
		ytDoubleArray *	chFreqs,
		int	size,
		int	num,
		int *	set,
		int	setSize,
		int	limit,
		int	resetPosition,
		ytIntArray *	W,
		ytDoubleArray *	P,
		ytRNG *	rng
	)

Samples nodes by random walk.

Parameters

g	global graph where we perform random walk.
size	total (max) number of set.
num	number of nodes to be sampled.
set
resetPosition	if 1, reset the node position when the new node is selected.
W	working array
P	working array

NNSR_shareNetwork()

void NNSR_shareNetwork	(	NNSRWork *	work,
		double	threshold
	)

Shares a global network by collecting edges among processors.

Parameters

work
threshold	cut off threshold for edges collected from processors.

---

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

• algo/NNSR.c