normal_cumulative.h File Reference

Normal cumulative sampler. More...

Go to the source code of this file.

Data Structures
struct	cfe_normal_cumulative

Typedefs
typedef struct cfe_normal_cumulative	cfe_normal_cumulative

Functions
void	cfe_normal_cumulative_init (cfe_normal_cumulative *s, mpf_t sigma, size_t n, bool two_sided)
void	cfe_normal_cumulative_free (cfe_normal_cumulative *s)
void	cfe_normal_cumulative_sample (mpz_t res, cfe_normal_cumulative *s)
void	cfe_normal_cumulative_precompute (cfe_normal_cumulative *s)
size_t	cfe_locate_int (mpz_t sample, size_t start, size_t end, cfe_vec *v, mpz_t middle_value)
void	cfe_normal_cumulative_sample_vec (cfe_vec *res, cfe_normal_cumulative *s)
void	cfe_normal_cumulative_sample_mat (cfe_mat *res, cfe_normal_cumulative *s)

Detailed Description

Normal cumulative sampler.

Typedef Documentation

cfe_normal_cumulative

typedef struct cfe_normal_cumulative cfe_normal_cumulative

cfe_normal_cumulative samples random values from the cumulative normal (Gaussian) probability distribution, centered on 0. This sampler is the fastest, but is limited only to cases when sigma is not too big, due to the sizes of the precomputed tables. Note that the sampler offers arbitrary precision but the implementation is not constant time.

Function Documentation

cfe_normal_cumulative_init()

void cfe_normal_cumulative_init	(	cfe_normal_cumulative *	s,
		mpf_t	sigma,
		size_t	n,
		bool	two_sided
	)

Initializes an instance of cfe_normal_cumulative sampler. It assumes mean = 0. Values are precomputed when this function is called, so that normal_cumulative_sample merely returns a precomputed value.

Parameters

s	A pointer to an uninitialized struct representing the sampler
sigma	Standard deviation
n	Precision parameter
two_sided	Boolean defining if we limit only to non-negative integers

cfe_normal_cumulative_free()

void cfe_normal_cumulative_free

(

cfe_normal_cumulative *

s

)

Frees the memory occupied by the struct members. It does not free memory occupied by the struct itself.

Parameters

s	A pointer to an instance of the sampler (initialized cfe_normal_cumulative struct)

cfe_normal_cumulative_sample()

void cfe_normal_cumulative_sample	(	mpz_t	res,
		cfe_normal_cumulative *	s
	)

Samples discrete cumulative distribution with precomputed values.

Parameters

res	The random number (result value will be stored here)
s	A pointer to an instance of the sampler (initialized cfe_normal_cumulative struct)

cfe_normal_cumulative_precompute()

void cfe_normal_cumulative_precompute

(

cfe_normal_cumulative *

s

)

Precomputes the values for sampling. This can be used only if sigma is not too big.

Parameters

s	A pointer to an instance of the sampler (initialized cfe_normal_cumulative struct)

cfe_locate_int()

size_t cfe_locate_int	(	mpz_t	sample,
		size_t	start,
		size_t	end,
		cfe_vec *	v,
		mpz_t	middle_value
	)

Given a random value from an interval and a cumulative distribution of mpz_t values, it returns a sample distributed accordingly using binary search

Parameters

sample	Target integer
start	Start index
end	End index
v	Vector of values
middle_value	Temporary placeholder for value comparison (to avoid unnecessary memory allocations)

Returns

index of target integer

cfe_normal_cumulative_sample_vec()

void cfe_normal_cumulative_sample_vec	(	cfe_vec *	res,
		cfe_normal_cumulative *	s
	)

Sets the elements of a vector to random numbers with the normal_cumulative sampler.

cfe_normal_cumulative_sample_mat()

void cfe_normal_cumulative_sample_mat	(	cfe_mat *	res,
		cfe_normal_cumulative *	s
	)

Sets the elements of a matrix to random numbers with the normal_cumulative sampler.