pargee: demonstrate a parallelized, out-of-memory based GEE solver

Description Usage Arguments Details Value Author(s) Examples

Description

demonstrate a parallelized, out-of-memory based GEE solver

Usage

1
pargee(gd, family, binit, maxit = 20, tol = 1e-06)

Arguments

gd

instance of gdManager-class

family

GLM stats family instance, e.g., gaussian

binit

initial value for regression parameter vector

maxit

maximum number of iterations allowed

tol

convergence criterion – will stop when largest maximum absolute size of any component of the iteration step is smaller than tol

Details

very preliminary demonstration, mildly validated on orthodontic data for gaussian case

Value

regression parameter vector

Author(s)

V Carey <stvjc@channing.harvard.edu>

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
## to obtain multicore performance, please do
# library(doMC)
# registerDoMC([num cores])  # give number of cores you want to use
## before running this example
 if (file.exists(fd <<- system.file("mgrs/flatOrth.rda", package="geeni"))) {
  load(fd)
  pargee(flatOrth, gaussian, c(0,0,0))
  } else {
  cat("you need to run the buildff vignette to update support information in the installed image of geeni\n")
  cat("try running Sweave(system.file('doc/buildff.Rnw',package='geeni')) and\n")
  cat("if all goes well, you can then run example(pargee)\n")
  }

geeni documentation built on May 2, 2019, 4:48 p.m.