Tutorial:Getting started: Compiling MSL with external libraries

From MSL-Libraries
Jump to navigationJump to search

Certain functions of MSL require the installation of some open-source external libraries.

  • GSL (GNU Scientific Library) [[1]]
  • BOOST (portable C++ source libraries) [[2]]
  • GLPK (GNU linear programming kit) [[3]]
  • R (R statistical and graphics project) [[4]]

What external libraries are really needed?

Only GSL is currently recommended and turned on by default in the Makefile

The other external libraries are needed only to support particular object or applications.


How to enable inclusion of the external libraries

By default MSL requires the GSL library and will be compiled _without_ the support of the other external libraries and some functionality won't be available. To include the libraries, you need to set some environmental variables.

If you use bash, add the following lines to your .bashrc

#  $MSL_GSL    set to "T" if GSL is installed or else to "F" (default is T)
#  $MSL_BOOST  set to "T" if BOOST is installed or else to "F" (default is F)
#  $MSL_GLPK   set to "T" if GLPK is installed or else to "F" (default is F)
#  $MSL_R  set to "T" if R is installed or else to "F" (default is F)
export MSL_GSL=T
export MSL_GLPK=F
export MSL_BOOST=F
export MSL_R=F
#export MSL_EXTERNAL_LIB_DIR=/usr/lib
#export MSL_EXTERNAL_INCLUDE_DIR=/usr/include
#export MSL_MACOS=T
#export MSL_DEBUG=F
#export MSL_DIR=$HOME/software/mslib

If you use tcsh, add the following lines to your .cshrc

#  $MSL_GSL    set to "T" if GSL is installed or else to "F" (default)
#  $MSL_BOOST  set to "T" if BOOST is installed or else to "F" (default)
#  $MSL_GLPK   set to "T" if GLPK is installed or else to "F" (default)
#  $MSL_R  set to "T" if R is installed or else to "F" (default)
setenv MSL_GSL T
setenv MSL_GLPK T
setenv MSL_BOOST T
setenv MSL_R T
#setenv MSL_EXTERNAL_LIB_DIR /usr/lib
#setenv MSL_EXTERNAL_INCLUDE_DIR /usr/include

The default location of the libraries is /usr/lib. If you have installed them in a custom location (such as that used below, ~/software/mslib/ext_libs) set the $MSL_EXTERNAL_LIB_DIR environmental variable to the correct location. You will also want to set the $MSL_EXTERNAL_LIB_INCLUDE_DIR if using a custom location (such as that used below, ~/software/mslib/ext_includes).

GSL

Many functions of MSL depend on [GNU Scientific library (GSL)], which is therefore highly recommended.

For example, GSL is required by the following objects:

If GSL is enabled, you should see -D__GSL_ in the compilation command when using make.


To compile GSL

GSL can be installed from the software repository (in Ubuntu, libgsl0-dev) or downloaded and installed from a tar file. For a more specific reference see the GSL homepage, but here is an example of the installation procedure:

>tar xzvf gsl-1.14.tar.gz 
>cd gsl-1.14
>./configure --prefix=/path/to/desired/directory
>make
>make check
>make install
>cp .libs/libgsl.a ~/software/mslib/ext_libs/
>cp cblas/.libs/libgslcblas.a ~/software/mslib/ext_libs/
  1. Note the -H, this should follow the symlinks, otherwise you get copies of links with relative paths and then you get errors making MSL that say gsl/gsl_rng.h not found, when you see it there!
> mkdir ~/software/mslib/ext_includes/gsl
> cp -H gsl/* ~/software/mslib/ext_includes/gsl

BOOST

The following libraries from the Boost C++ Libraries are used in some MSL objects and programs. In Ubuntu, it is available in the package libboost-all-dev.

Note: if boost is included, version 1.40 (???) or later is required.

If BOOST is enabled, you should see -D__BOOST__ -DBOOST_DISABLE_THREADS in the compilation command when using make.

To Compile BOOST

>tar xzvf boost-VERSION.tar.gz
>cd boost-VERSION
>./bootstrap.sh --prefix=~/software/mslib/ext_libs
>./b2
> cp ./stage/lib/libboost_*.a ~/software/mslib/ext_libs
> cp -r boost ~/software/mslib/ext_includes/

Assuming MSL library is installed in ~/software/mslib/. You will need to make the ext_includes and ext_libs directories before copying.

GLPK

The GNU linear programming kit is required by the LinearProgrammingOptimization object.

If GLPK is enabled, you should see -D__GLPK__ in the compilation command when using make.

./configure --prefix=/usr/lib --disable-dl --disable-odbc --disable-mysql --enable-static --without-gmp --without-zlib
 make
 make install
 cp /usr/lib/libglpk.a MSLHOME/ext_libs/
 cp /usr/include/glpk.h MSLHOME/ext_includes/

NOTE: the current libglpk in Ubuntu appears to be incompatible with MSL

R

The R project is used by the getDihedrals program. In Ubuntu the package is r-base.

If R is enabled, you should see -D__R__ in the compilation command when using make.

To Compile R

Compiling R on MAC OS

Install R and copy headers into MSL_EXTERNAL_INCLUDE_DIR

 cp -r /Library/Frameworks/R.framework/Resources/include/ $MSL_EXTERNAL_INCLUDE_DIR

Download Rcpp

R CMD INSTALL Rcpp_0.8.5.tar.gz
cp /Library/Frameworks/R.framework/Resources/library/Rcpp/lib/x86_64/libRcpp.a $MSL_EXTERNAL_LIB_DIR/
cp -r /Library/Frameworks/R.framework/Resources/library/Rcpp/include $MSL_EXTERNAL_INCLUDE_DIR/
R CMD INSTALL RInside_0.2.3.tar.gz
cp /Library/Frameworks/R.framework/Resources/library/RInside/lib/x86_64/libRInside.a $MSL_EXTERNAL_LIB_DIR/
cp -r /Library/Frameworks/R.framework/Resources/library/RInside/include $MSL_EXTERNAL_INCLUDE_DIR/


If the above install does not work, one can install from source here is a good resource:

http://cran.r-project.org/bin/macosx/RMacOSX-FAQ.html#Quick-start
Compiling R on LINUX OS

Success using shared libraries and compiling MSL as shared lib, but static lib has not been done yet. I will link my Makefile for shared-R compile once its cleaned up a bit

First try using the 'apt' system in linux:

sudo apt-get install r-base

If the 'apt' version is not the latest version that you need, then install latest version of R:

sudo apt-get build-dep r-base
tar xvzf R-2.11.1.tar.gz
cd R-2.11.1
./configure
make
sudo make install


Install Rcpp:

sudo /usr/local/bin/R CMD INSTALL Rcpp_0.8.5.tar.gz
cp /Library/Frameworks/R.framework/Resources/library/Rcpp/lib/x86_64/libRcpp.a $MSL_EXTERNAL_LIB_DIR/
cp -r /Library/Frameworks/R.framework/Resources/library/Rcpp/include $MSL_EXTERNAL_INCLUDE_DIR/


Install RInside:

R CMD INSTALL RInside_0.2.3.tar.gz

cp /Library/Frameworks/R.framework/Resources/library/RInside/lib/x86_64/libRInside.a $MSL_EXTERNAL_LIB_DIR/
cp -r /Library/Frameworks/R.framework/Resources/library/RInside/include $MSL_EXTERNAL_INCLUDE_DIR/
 

Copy headers into MSL_EXTERNAL_INCLUDE_DIR

 cp -r /Library/Frameworks/R.framework/Resources/include/ $MSL_EXTERNAL_INCLUDE_DIR