This page contains instructions to download and install the R-INLA package

Before you start, make sure you have installed the package 'sp' as INLA requires it. You may also install the packages: numDeriv, Rgraphviz (from Bioconductor), fields, rgl, mvtnorm, multicore, pixmap, but these are required only for advanced use and are normally not required and are hence only 'suggested'.

You can then install the R-INLA package using one of the following alternatives:

1. The new way (from 3rd January 2014)

We have created a standard R-repository, so that 'install.packages' and 'update.packages' will work as expected. You need to add the address to the INLA-repository, as

> install.packages("INLA", repos="")

for the stable version, OR

> install.packages("INLA", repos="")

for the testing version. Do simular using 'update.packages()'.

You can also just append the INLA-repos to the  global 'repos' variable, like

> options(respos = c(getOption("repos"), INLA=""))

and then you can do

> update.packages("INLA")

etc, to install and update the package, without specifying the repos.  After a testing-period, this will be the only
way to install the package and option #2 and #3 below, will be disabled.

The easy way

  There are two versions of the R-INLA package; one that is updated 'once in a while' and one that is in active development (the testing-version).  
  To install either one of these, type one of the following command line in R
> source("")

## or
> source("")
  to install the more stable version or the testing version. You can later upgrade the package, using either
   > inla.upgrade()  
## or
> inla.upgrade(testing=TRUE)

If you have installed the testing version, then 'inla.upgrade()' will take to back to the more stable version, and if you have installed the more
stable version, then 'inla.upgrade(testing=TRUE)' will upgrade you to the most recent testing version. (See this link for further information.)

3. The regular way:

Download the (more stable) package choosing one of the following alternatives


and install it. For Linux and Mac, do

> install.packages("INLA.tgz", repos=NULL, type="source")

for Windows, start R and select the Packages menu, then Install
package from local zip file by find and highlight the location of the
zip file and click on open. Upgrade the package as described above.

Small prints:
  • All source code for this project is now hosted by google code and you can view the latest changes here.
  • Software licenses
  • In case you want to know the details the inla-program which is the backend-program which do all the computations, you can download the manual here. However, it is not up-to-date with respect to new models but the input/output file-format is well described. 
  • Compile it yourself? Download build-packages and compile it yourself. You will need to install mercurial which is available in all larger Linux distros and binaries for Mac OSX is available from the same site.