In ruaridhw/r-AMPL: AMPL to R API

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

rAMPL

Provides convenience wrappers around the Python package amplpy using reticulate.

Installation

You can install AMPL from GitHub with:

# install.packages("devtools")
devtools::install_github("ruaridhw/r-AMPL")

As this package relies on the amplpy API you will need a Python installation and the required package.

You can check if you have Python installed by opening a Terminal and running:

python -V

If this doesn't print a Python version, you will need to install it first and then run:

```{bash pip-install, eval=FALSE} pip install amplpy

You can obviously run this with Anaconda or a Python installation other
than the system default.

## Usage

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

First, start an AMPL environment. The default path is `/Applications/ampl`
and assumes you have an AMPL binary installed here.

```r
library(AMPL)
ampl <- ampl_env()

Next, we will point to the location of the model and data files. This package includes the AMPL book's example models so we'll use its "steel" model:

model_file <- system.file("models/steel.mod", package = "AMPL")
data_file <- system.file("models/steel.dat", package = "AMPL")

Finally, run the model:

ampl <- set_solver(ampl, "/Applications/ampl/cplex")
ampl <- read_model(ampl, model_file)
ampl <- read_data(ampl, data_file)
ampl <- solve_model(ampl)
ampl$close()

Using the magrittr pipe to chain commands, this can be executed without continuously repeating assignment:

ampl <- ampl_env() %>%
  set_solver("/Applications/ampl/cplex") %>%
  read_model(model_file) %>%
  read_data(data_file) %>%
  solve_model()
steel_results <- get_data_all(ampl)
ampl$close()

steel_results$Parameters$market
steel_results$Variables
steel_results$Objectives

Additional Info

Configure Python

If you installed amplpy using a Python installation other than the system default, such as conda, you can specify which version to use with a command similar to the following prior to loading AMPL:

reticulate::use_python("/Users/username/anaconda3/envs/my_env/bin/python")

Configure AMPL

The default ampl binary location is /Applications/ampl however if you have AMPLDev installed, you can use this binary instead:

ampl <- ampl_env("/Applications/AMPLDev.app/Contents/Resources/AMPL")

If you do this, there are a couple of steps to be aware of:

The AMPLDev binary must be licensed correctly otherwise R will crash. (In the future, this package will check this ahead of time and throw a warning). You can check this yourself by running the following in a Terminal:

{bash} cd /Applications/AMPLDev.app/Contents/Resources/AMPL ampl --version

You cannot shortcut this step without the cd command due to OptiRisk's License Manager.

If this doesn't work, ensure that the file LICENSE.KEY appears in both the Resources and AMPL directories.
Any paths inside your AMPL model files will need to be absolute paths rather than relative paths. The AMPL environment runs from inside the AMPLDev app so relative file paths will be evaluated against this directory which is almost certainly not the user's intention.