rsparsematrix function

Random Sparse Matrix

Generate a random sparse matrix efficiently. The default has rounded gaussian non-zero entries, and rand.x = NULL generates random pattern matrices, i.e. inheriting from nsparseMatrix.

rsparsematrix(nrow, ncol, density, nnz = round(density * maxE),
              symmetric = FALSE,
              rand.x = function(n) signif(rnorm(n), 2), ...)

Arguments

• nrow, ncol: number of rows and columns, i.e., the matrix dimension (dim).
• density: optional number in $[0,1]$, the density is the proportion of non-zero entries among all matrix entries. If specified it determines the default for nnz, otherwise nnz needs to be specified.
• nnz: number of non-zero entries, for a sparse matrix typically considerably smaller than nrow*ncol. Must be specified if density is not.
• symmetric: logical indicating if result should be a matrix of class symmetricMatrix. Note that in the symmetric case, nnz denotes the number of non zero entries of the upper (or lower) part of the matrix, including the diagonal.
• rand.x: NULL or the random number generator for the x slot, a function such that rand.x(n) generates a numeric vector of length n. Typical examples are rand.x = rnorm, or rand.x = runif; the default is nice for didactical purposes.
• ...: optionally further arguments passed to sparseMatrix(), notably repr.

Details

The algorithm first samples encoded $(i,j)$s without replacement, via one dimensional indices, if not symmetric

sample.int(nrow*ncol, nnz), then---if rand.x is not NULL---gets x <- rand.x(nnz) and calls sparseMatrix(i=i, j=j, x=x, ..). When rand.x=NULL, sparseMatrix(i=i, j=j, ..) will return a pattern matrix (i.e., inheriting from nsparseMatrix).

Returns

a sparseMatrix, say M of dimension (nrow, ncol), i.e., with dim(M) == c(nrow, ncol), if symmetric

is not true, with nzM <- nnzero(M) fulfilling nzM <= nnz and typically, nzM == nnz.

Author(s)

Martin Maechler

Examples

set.seed(17)# to be reproducible
M <- rsparsematrix(8, 12, nnz = 30) # small example, not very sparse
M
M1 <- rsparsematrix(1000, 20,  nnz = 123,  rand.x = runif)
summary(M1)

## a random *symmetric* Matrix
(S9 <- rsparsematrix(9, 9, nnz = 10, symmetric=TRUE)) # dsCMatrix
nnzero(S9)# ~ 20: as 'nnz' only counts one "triangle"

## a random patter*n* aka boolean Matrix (no 'x' slot):
(n7 <- rsparsematrix(5, 12, nnz = 10, rand.x = NULL))

## a [T]riplet representation sparseMatrix:
T2 <- rsparsematrix(40, 12, nnz = 99, repr = "T")
head(T2)