Armadillo: C++ library for linear algebra & scientific computing (original) (raw)

Stable Version

• armadillo-14.4.3.tar.xz

Related Software

• mlpack - extensive library of machine learning algorithms
• ensmallen - C++ library for efficient numerical optimisation
• CARMA - interface between Armadillo and Python / NumPy

Installation Notes

• See the README file in the .tar.xz package for full installation instructions

• If you encounter any bugs or regressions, please report them

• If you use Armadillo in your research and/or software, please cite the associated papers;
  citations are useful for the continued development and maintenance of the library

• Linux based operating systems (eg. Fedora, Ubuntu, Red Hat, SUSE, Debian, etc)

  • Before installing Armadillo, first install OpenBLAS and LAPACK, along with the corresponding development/header files
  • Recommended packages to install before installing Armadillo:
    * Fedora & Red Hat: cmake, openblas-devel, lapack-devel, arpack-devel, SuperLU-devel
    * Ubuntu & Debian: cmake, libopenblas-dev, liblapack-dev, libarpack2-dev, libsuperlu-dev
  • Pre-built Armadillo packages are provided by many Linux-based operating systems:Fedora,Debian,Ubuntu,openSUSE,Arch
    the pre-built packages may not be the latest version; if you're encountering problems, use the official stable version provided here

• macOS

  • The "Accelerate" framework is used for accessing BLAS and LAPACK functions; see the README file in the package for more information
  • Pre-built Armadillo packages can also be installed viaMacPorts orHomebrew
    the pre-built packages may not be the latest version; if you're encountering problems, use the official stable version provided here

• Windows

  • The Armadillo package contains pre-compiled OpenBLAS as well as MSVC project files to compile the example program, tested on Windows 10 (64 bit) with Visual Studio 2019;
    you may need to make adaptations for later versions of Windows and/or the compiler

  • Intel MKL can also be used as a fast implementation of BLAS and LAPACK

  • Caveat: for any high performance scientific/engineering workloads, we strongly recommend using a Linux based operating system

• Cross-Platform

  • Armadillo packages for the Conda package management system (part of Anaconda):
    * conda-forge Armadillo package
    * conda-forge Armadillo feedstock

  • Armadillo recipe for the Conan package management system:
    * Armadillo recipe
    * source for Armadillo recipe