How to setup R and JAGS on Amazon EC2

Yesterday I setup a Amazon EC2 machine and installed R, JAGS and rjags. Unfortunately I was unable to compile the latest version of JAGS. The configuration step failed with an error that says that a specific LAPACK library is missing. I tried to install the missing library manually but for some reason JAGS doesn’t recognize the new version. After an hour, I decided to go with the older JAGS version. Here is how I installed all three packages.

[If you never worked on an EC2 machine: Here are two [1,2] helpful tutorials how start an EC2 instance and install the command line tool on your computer to access it via ssh.]

Login via ssh. In a first step we install an older version of R. It is necessary to use the older version in order to get JAGS 1.0.4 running later. After downloading the R source file, we unzip it, cd into the folder and start installing a couple of additional libraries and compilers that we need to compile R (and later JAGS). We then configure the R installer without any x-window support and install everything. The last line makes sure that you can start R with simply typing ‘R’.

cd ~
wget http://cran.r-project.org/src/base/R-2/R-2.10.0.tar.gz 
tar xf R-2.10.0.tar.gz
cd ~/R-2.10.0
sudo yum install gcc
sudo yum install gcc-c++
sudo yum install gcc-gfortran
sudo yum install readline-devel
sudo yum install make
./configure --with-x=no
make
PATH=$PATH:~/download/R-2.10.0/bin/

Next, try to start R and install the coda package.

R
install.packages('coda')
q()

Now, we proceed with installing JAGS. First, get some libraries and the JAGS source code. Unzip, cd into the folder and run configure with specifying the shared library folder. The specification of the shared library path is crucial, otherwise rjags fails to load in R.

sudo yum install blas-devel lapack-devel
cd ~/download/
wget http://sourceforge.net/projects/mcmc-jags/files/JAGS/1.0/JAGS-1.0.4.tar.gz/download
tar xf JAGS-1.0.4.tar.gz
cd ~/download/JAGS-1.0.4
./configure --libdir=/usr/local/lib64 
make 
sudo make install

In a last step, download the appropriate rjags version and install it it.

cd ~/download/
wget http://cran.r-project.org/src/contrib/Archive/rjags/rjags_1.0.3-13.tar.gz
R --with-jags-modules=/usr/local/lib/JAGS/modules/ CMD INSTALL rjags_1.0.3-13.tar.gz 

Sources of inspiration. Here, here, here and here.