qlcMatrix (version 0.9.2)

rSparseMatrix: Construct a random sparse matrix

Description

This convenience function constructs a random sparse matrix of specified size, with specified sparsity. This is mainly useful for testing speed and memory load of sparse matrix manipulations

Usage

rSparseMatrix(nrow, ncol, nnz, 
	rand.x = function(n) round(rnorm(nnz), 2), ...)

Value

Returns a sparse matrix of the type dgCMatrix. Defaults to random numeric entries with two decimal digits, generated randomly from a normal distribution with mean = 0 and sd = 1.

When rand.x = NULL then the result is a pattern matrix of type ngCMatrix.

Details

The sparsity of the resulting matrix (i.e. the fraction of non-zero entries to all entries) is $\frac{nnz}{nrow * ncol}$.

See Also

For random permutation matrices, see pMatrix-class. Specifically note the construction option (p10 <- as(sample(10),"pMatrix")).

Examples

# example with reasonably large (100.000 by 100.000) but sparse matrix
# (only one in 10.000 entries is non-zero). On average 10 entries per column.
X <- rSparseMatrix(1e5, 1e5, 1e6)
print(object.size(X), units = "auto")

# speed of cosine similarity
system.time(M <- cosSparse(X))

# reduce memory footprint by removing low values
print(object.size(M), units = "auto")
M <- drop0(M, tol = 0.1)
print(object.size(M), units = "auto")