Hardware-Accelerated SVD via LAPACK — accelerate

Uses R's linked LAPACK (which on macOS is Apple Accelerate/AMX, on Linux is typically OpenBLAS or MKL) for hardware-optimised SVD computation.

Usage

accelerate_svd(A, thin = TRUE)

Arguments

A: Matrix (m x n). Input matrix.
thin: Logical. If TRUE (default), compute thin SVD.

Value

A list with components u, d and v, so that A = u %*% diag(d) %*% t(v). With thin = TRUE, u is m x k and v is n x k, where k = min(m, n). With thin = FALSE, u is m x m and v is n x n, and only their first k columns pair with d.

Examples

set.seed(1L)
A <- matrix(rnorm(8 * 5), nrow = 8, ncol = 5)
res <- accelerate_svd(A, thin = TRUE)
length(res$d)
#> [1] 5
all.equal(res$d, svd(A)$d)
#> [1] TRUE
max(abs(A - res$u %*% diag(res$d) %*% t(res$v)))
#> [1] 7.271961e-15