notes/profiling.R

In pknight24/bema: Bulk Eigenvalue Matching Analysis

library(microbenchmark)
library(profvis)
library(irlba)

n <- 100
p <- 500
K <- 5
rho <- 5
mu_k <- rho * sqrt(p/n)
vectors <- svd(matrix(rnorm(p*p), p, p))$u[,1:K]
theta <- 5
sigma2 <- 1
Sigma <- tcrossprod(mu_k * vectors) + diag(rgamma(p, theta, rate = theta/sigma2))

X <- mvrnorm(n = n, mu = rep(0, p), Sigma = Sigma)
S <- 1/n * tcrossprod(X)
step1_out <- step1(eigen(S)$values, n, p, 0.1, B = 20)
profvis({
  step2(eigen(S)$values, n, p, beta = 0.2, M = 500, step1_output = step1_out, use_irlba = TRUE)
  step2(eigen(S)$values, n, p, beta = 0.2, M = 500, step1_output = step1_out, use_irlba = FALSE)
})

microbenchmark(svd(X, nu = 0, nv = 0),
               irlba(X, nu = 1, nv = 0))