Using Microsoft R Open with conda¶
NOTE: December 12, 2017 update: The packages in the MRO channel were last updated in 2016 and are now deprecated. They have been copied to https://repo.continuum.io/pkgs/mro-archive and in about one week they will be removed from their current locations at https://repo.continuum.io/pkgs/mro/ and https://anaconda.org/mro . Soon after that the new versions of the MRO packages will be posted.
- 64-bit systems only for all operating systems—Windows, macOS and Linux.
- Windows 7.0 SP1, Windows 8.1, Windows 10, Windows Server 2008 R2 SP1 and Windows Server 2012.
- macOS—Mavericks (10.9), Yosemite (10.10) and El Capitan (10.11).
- Linux—CentOS, Red Hat Enterprise Linux, Debian and Ubuntu.
You can install MRO and other R language packages several ways:
- Add the MRO channel to your .condarc configuration file above the R
channel, and then use
conda install r. Use conda to install other packages.
- Specify the MRO channel on the command line each time you use
conda install -c mro r.
- Install R packages from the Comprehensive R Archive Network (CRAN) or the Microsoft R Application Network (MRAN). These R packages install into the currently active conda environment. MRO includes the checkpoint package, which installs from MRAN and is designed to offer enhanced reproducibility.
To use MRO or R language packages, first activate the conda environment where they are installed to set your environment variables properly.
NOTE: Errors may result if you try to execute a program at the path name in that environment without first activating the environment.
Each conda environment may have packages installed from the channel “r” or the channel “mro,” but no conda environment should contain packages from both channels. Doing so may result in errors.
Handle these errors in one of the following ways:
- Use the command
conda removeto remove the packages that were installed incorrectly.
- Create a new environment, and then install the correct packages there.
The Intel Math Kernel Library (MKL) extensions are available for MRO on Windows and Linux, while the Accelerate library is used instead on macOS.
NOTE: By using our install process, you are certifying that you have agreed to the MKL license. Proceeding further is an implicit agreement to this license.
When MKL is not installed, each time the R interpreter starts it displays a message such as:
No performance acceleration libraries were detected. To take advantage of the available processing power, also install MKL for MRO 3.2.3. Visit http://go.microsoft.com/fwlink/?LinkID=698301 for more details.
When the MKL extensions are properly installed, each time the R interpreter starts it displays a message such as:
Multithreaded BLAS/LAPACK libraries detected.
NOTE: BLAS/LAPACK libraries such as MKL are implementations of the Basic Linear Algebra Subprograms specification of the LAPACK Linear Algebra PACKage.
Installing Windows MKL¶
Download the proper MKL installer file.
Find the directory where the R language is located. It is a directory called
Rin one of the following locations:
- The current user’s home directory.
- The directory for the current environment.
EXAMPLE: An Anaconda user on Windows 8 with the user name
jsmithand an environment named
Run the installer.
When prompted to “Enter a path to MRO,” enter the path.
Starting the R language interpreter now displays the message showing that MKL is enabled.
If you wish to uninstall MKL on Windows and continue using the R language, uninstall MKL from the Windows Control Panel.
Installing Linux MKL¶
Download and extract the proper MKL package to a temporary folder with a name such as
Rdirectory. The path has the form:
In the remaining steps, the path to the
Rdirectory is represented by $PREFIX.
Ensure that you have write permissions to $PREFIX.
Make backups of
Copy all of the
$PREFIX/lib/R/lib. This may prompt you to overwrite
$PREFIX/lib/R/etc/Rprofile.siteto add the following 2 lines to the top:
Run this command:
R CMD INSTALL $REVOMATH/RevoUtilsMath.tar.gz