tatami_r Namespace Reference

tatami bindings for arbitrary R matrices. More...

Classes
class	UnknownMatrix
	Unknown matrix-like object in R. More...
struct	UnknownMatrixOptions
	Options for data extraction from an UnknownMatrix. More...

Functions
manticore::Executor &	executor ()
void	set_executor (manticore::Executor *ptr)
template<class Function_ , class Index_ >
int	parallelize (const Function_ fun, const Index_ tasks, int threads)
template<class Function_ >
void	parse_SVT_SparseMatrix (const Rcpp::RObject &matrix, const Function_ fun)

Detailed Description

tatami bindings for arbitrary R matrices.

Function Documentation

executor()

manticore::Executor & tatami_r::executor ( )

inline

Retrieve a global manticore::Executor object for all tatami_r applications. This function is only available if TATAMI_R_PARALLELIZE_UNKNOWN is defined.

Returns

Reference to a global manticore::Executor. If set_executor() was called with a non-NULL pointer, the provided instance will be used; otherwise, a default instance will be instantiated.

parallelize()

template<class Function_ , class Index_ >

int tatami_r::parallelize	(	const Function_	fun,
		const Index_	tasks,
		int	threads )

Template Parameters

Function_	Function to be executed.
Index_	Integer type for the task indices.

Parameters

fun	Function to run in each thread. This is a lambda that should accept three arguments: Integer containing the thread ID in [0, threads). Integer specifying the index of the first task to be executed in a thread. This will lie in [0, tasks). Integer specifying the number of tasks to be executed in a thread. This will lie in (0, tasks), i.e., it is always positive.
tasks	Number of tasks to be executed. This should be non-negative.
threads	Number of threads to parallelize over. This should be positive.

This function is a drop-in replacement for tatami::parallelize(). The series of integers from [0, tasks) is deterministically split into K non-overlapping non-empty contiguous ranges where K <= threads. Each range is passed to fun for parallel execution with the standard <thread> library. Serialization can be achieved via <mutex> in most cases, or manticore::Executor::run() if the task must be performed on the main thread (see executor()).

This function is only available if TATAMI_R_PARALLELIZE_UNKNOWN is defined.

Returns

The number of workers (K) that were actually used. K is guaranteed to be no greater than threads (or 1, if the latter is not positive). fun() will have been called once for each of the thread IDs [0, ..., K - 1].

parse_SVT_SparseMatrix()

template<class Function_ >

void tatami_r::parse_SVT_SparseMatrix	(	const Rcpp::RObject &	matrix,
		const Function_	fun )

Parse the contents of a SVT_SparseMatrix from the DelayedArray package. This accounts for different versions of the class definition, different types of the values, and the presence of lacunar leaf nodes.

Template Parameters

Function_

Function to be applied at each leaf node.

Parameters

matrix

The SVT_SparseMatrix object.

fun

Function to apply to each leaf node, accepting four arguments: c, an integer specifying the index of the leaf node, i.e., the column index. indices, an Rcpp::IntegerVector containing the sorted, zero-based indices of the structural non-zero elements in this node (i.e., column). all_ones, a boolean indicating whether all values in this node/column are equal to 1. values, an Rcpp::IntegerVector, Rcpp::LogicalVector or Rcpp::NumericVector containing the values of the structural non-zeros. This should be of the same length as indices. It should be ignored if all_ones = true. The return value of this function is ignored. Note that fun may not be called for all c - if leaf nodes do not contain any data, they will be skipped.

set_executor()

void tatami_r::set_executor ( manticore::Executor * ptr )

inline

Set a global manticore::Executor object for all tatami_r applications. This function is only available if TATAMI_R_PARALLELIZE_UNKNOWN is defined. Calling this function is occasionally necessary if executor() resolves to different instances of a manticore::Executor across different libraries.

Parameters

Pointer

to a global manticore::Executor, or NULL to unset this pointer.