Parallel Linear Mixed Model

Embarrassingly Parallel Linear Mixed Model calculations spread across local cores which repeat until convergence.


Travis-CI Build Status CRAN_Status_Badge

The goal of lmmpar is to ...

Installation

You can install lmmpar from github with:

# install.packages("devtools")
devtools::install_github("fulyagokalp/lmmpar")

Example

This is a basic example which shows you how to solve a common problem:

# Set up fake data
n <- 10000  # number of subjects
m <- 4      # number of repeats
N <- n * m  # true size of data
p <- 50     # number of betas
q <- 2      # width of random effects
 
# Initial parameters
# beta has a 1 for the first value.  all other values are ~N(10, 1)
beta <- rbind(1, matrix(rnorm(p, 10), p, 1))
R <- diag(m)
D <- matrix(c(16, 0, 0, 0.025), nrow = q)
sigma <- 1
 
# Set up data
subject <- rep(1:n, each = m)
repeats <- rep(1:m, n)
 
subj_x <- lapply(1:n, function(i) cbind(1, matrix(rnorm(m * p), nrow = m)))
X <- do.call(rbind, subj_x)
Z <- X[, 1:q]
subj_beta <- lapply(1:n, function(i) mnormt::rmnorm(1, rep(0, q), D))
subj_err <- lapply(1:n, function(i) mnormt::rmnorm(1, rep(0, m), sigma * R))
 
# create a known response
subj_y <- lapply(
   seq_len(n),
   function(i) {
     (subj_x[[i]] %*% beta) +
       (subj_x[[i]][, 1:q] %*% subj_beta[[i]]) +
       subj_err[[i]]
   }
)
Y <- do.call(rbind, subj_y)
 
# run the algorithm in parallel to recover the known betas
ans <- lmmpar(
   Y,
   X,
   Z,
   subject,
   beta = beta,
   R = R,
   D = D,
   cores = 4,
   sigma = sigma,
   verbose = TRUE
)
 

News

lmmpar 0.1.0

  • New release

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("lmmpar")

0.1.0 by Fulya Gokalp Yavuz, 2 years ago


https://github.com/fulyagokalp/lmmpar


Report a bug at https://github.com/fulyagokalp/lmmpar/issues


Browse source code at https://github.com/cran/lmmpar


Authors: Fulya Gokalp Yavuz [aut, cre] , Barret Schloerke [aut]


Documentation:   PDF Manual  


MIT + file LICENSE license


Imports MASS, matrixcalc, mnormt, plyr, doParallel, bigmemory

Suggests testthat


See at CRAN