knn_parallel: Find K nearest neighbours for multiple query points

In bigutilsr: Utility Functions for Large-scale Data

Description

Find K nearest neighbours for multiple query points

Usage

knn_parallel(data, query = data, k, ..., ncores = bigparallelr::nb_cores())

Arguments

data	Mxd matrix of M target points with dimension d
query	Nxd matrix of N query points with dimension d (nb data and query must have same dimension). If missing defaults to data i.e. a self-query.
k	an integer number of nearest neighbours to find
...	Arguments passed on to nabor::knn eps An approximate error bound. The default of 0 implies exact matching. searchtype A character vector or integer indicating the search type. The default value of 1L is equivalent to "auto". See details. radius Maximum radius search bound. The default of 0 implies no radius bound.
ncores	Number of cores to use. Default uses bigparallelr::nb_cores().

Value

A list with elements nn.idx (1-indexed indices) and nn.dists (distances), both of which are N x k matrices. See details for the results obtained with1 invalid inputs.

Examples

## Not run: knn_parallel(matrix(1:4, 2), k = 2, ncores = 2)

bigutilsr documentation built on April 14, 2021, 1:06 a.m.