Skip to content

Latest commit

 

History

History
95 lines (68 loc) · 3.6 KB

INSTALL.md

File metadata and controls

95 lines (68 loc) · 3.6 KB

Downloading and installing the package

The package is primarily distributed in the form of source code and is hosted at http://github.com/slra/slra/. For some platforms, precompiled binaries are also available in the repository.

The package can be downloaded from http://slra.github.io/software.html, or directly from http://github.com/slra/slra/.

If the precompiled binaries are available for you platform, you should

  • In MATLAB/Octave: add the whole directory to the MATLAB path with the addpath command

  • In R (for Windows): run

    install.packages(repos=NULL, pkgs="Rslra_x.x.xxx");

    in the R console launched in the same directory where the package is.

Otherwise, please follow instructions in section Installing the package from source

Installing the package from source

Prerequisites

GSL library should be installed (if not using a precompiled Windows package for R). To use MATLAB/Octave or R interfaces, the corresponding environments should also be installed.

SLRA package uses LAPACK and BLAS libraries, which are included in MATLAB, Octave and R installations, so if you have MATLAB, Octave or R installed, you don't need to install LAPACK and BLAS libraries. The package has an optional binding to SLICOT library, which can speed up computations in some cases.

The source files of the libraries can be obtained at

Installation from source

  1. Get source

    Download from http://github.com/slra/slra/ and unpack to a directory

  2. Compile

    • type make matlab to produce a MEX binary file for MATLAB
    • type make matlab-win to produce a MEX binary file in Windows
    • type make octave to produce a MEX binary file for Octave
    • type make R to produce an R package and install it

  3. Install

    • In MATLAB/Octave you should add the whole directory to the MATLAB path with the addpath command

    • The R package should be loaded each time before using it by typing

      library(Rslra);

      in the R console.

Special instructions for Windows MEX binary file compilation

The target matlab-win uses the MinGW64 compiler. MinGW64 can be installed from http://mingw-w64.sourceforge.net/ or from http://cran.r-project.org/bin/windows/Rtools/. MSYS can be installed from http://www.mingw.org/wiki/MSYS.

Prior to running make matlab-win you should

  • compile GSL from source using the MSYS console
  • copy the gsl subfolder (with header files) to one of the include paths of the MinGW compiler
  • copy the libgsl.a and libgslcblas.a files to mex directory
  • set up the paths to MinGW and MATLAB in makefile

Notes for advanced users

The package contains a demo C++ program, which gives an example of using C++ interface and tests various SLRA problems (make testc). Documentation for C++ interface can be obtained by running Doxygen in cpp subdirectory.

Advanced compilation options can be found in other targets of makefile, but not all of targets may run on your machine "as is".

If you wish to try the SLICOT library, download it and copy to SLICOT subdirectory. You can copy only a few needed files (see makefile). Use make xxx-slicot-xxx.

Depending on the version of MATLAB, static binding to non-default BLAS and LAPACK (for example, ATLAS) can be faster. Use xxx-static target as a base for your compilation instructions.