sandbox/cent_test.R

In benkeser/haltmle.sim: Simulation study for HAL-TMLE

#! /usr/bin/env Rscript 
# This file was used to submit the simulation files to 
# a slurm-based Unix system. Using the sce.sh shell script
# one can submit each simulation in sequence. First, data files are
# created for each simulation. Those data files are then analyzed 
# in the 'run' execution. Then the results are collated in the 'merge'
# execution. 

# get environment variables
MYSCRATCH <- Sys.getenv('MYSCRATCH')
RESULTDIR <- Sys.getenv('RESULTDIR')
STEPSIZE <- as.numeric(Sys.getenv('STEPSIZE'))
TASKID <- as.numeric(Sys.getenv('SLURM_ARRAY_TASK_ID'))

# set defaults if nothing comes from environment variables
MYSCRATCH[is.na(MYSCRATCH)] <- '.'
RESULTDIR[is.na(RESULTDIR)] <- '.'
STEPSIZE[is.na(STEPSIZE)] <- 1
TASKID[is.na(TASKID)] <- 0

# get command lines arguments
args <- commandArgs(trailingOnly = TRUE)
if(length(args) < 1){
  stop("Not enough arguments. Please use args 'listsize', 'prepare', 'run <itemsize>' or 'merge'")
}

# get the list size #########
if (args[1] == 'listsize') {
  cat(1)
  # for testing, useful to just run the first job
  # to make sure everything saves correctly
  # cat(1)
}

# execute prepare job ##################
if (args[1] == 'prepare') {
  print("Yay preparing!")
}

# execute parallel job #################################################
if (args[1] == 'run') {
  if (length(args) < 2) {
    stop("Not enough arguments. 'run' needs a second argument 'id'")
  }
  id <- as.numeric(args[2])
  print(paste(Sys.time(), "arrid:" , id, "TASKID:",
              TASKID, "STEPSIZE:", STEPSIZE))
  for (i in (id+TASKID):(id+TASKID+STEPSIZE-1)) {
    print("Yay running!")
  }
}

# merge job ###########################
if (args[1] == 'merge') {
  print("Yay merging!")
}