qm

This repository contains the R client for QMachine, a web service for distributing computations. The client will be published as its own R package on CRAN -- hopefully as qm -- in the near future, but it is still very rough. In the meantime, you can install it and try it out thanks to the excellent devtools package:

library(devtools)
devtools::install_github('qmachine/qm-r')

Introduction

Having installed qm, you can now start distributing computations and also volunteering your own computer as part of a crowdsourced supercomputer, using only R. The convention used in this library is to transform input data x to output data y with a function f.

To volunteer to run jobs for the "test-from-r" box (more on this later),

library(qm)

while (TRUE) {
  qm::volunteer(box = 'test-from-r')
  Sys.sleep(1)
}

Some examples for submitting jobs are shown below, again using the "test-from-r" box. Note that, if your box has no volunteers, your jobs will not be executed.

library(qm)

qm_box <- 'test-from-r'

# Example 1: Computing `2 + 2` remotely

f <- function(x) x + 2
x <- 2
y <- qm::submit(box = qm_box, f = f, x = x)

print(y)

# Example 2: Summing numbers remotely with built-in (native) functions

print(qm::submit(box = qm_box, f = sum, x = 1:5))

The browser client developed for the original paper also provided higher-order map, reduce, and mapreduce patterns, but those have not been included in this package.

---