Linux/Unix Operating System

Installation of SVL in a general Linux/Unix operating system requires setting up your machine to handle C++ and Python. This means that an appropriate C++ compiler is required as well as some high-performance numerical algebra library packages must be installed.

• The SVL Pre-Analysis requires the following python3 libraries:

  • Numpy
  • Scipy
  • Matplotlib
  • Json

  These libraries can be installed using pip. For Linux/Unix systems, repositories can be used as well. In the case of Debian-based system use apt-get, for Fedora-based system use yum, and for MacOS use brew.

• The SVL Run-Analysis requires to download Eigen MUMPS and PETSc libraries.

  • The Eigen C++ library can be downloaded from this website.
    This package needs to be unzip and its content move (for instance) to /usr/include/eigen.
  • The MUMPS library can be downloaded from this website.
    This package needs to be unzip and compiled (for instance) at /usr/include/mumps.
  • The Pestc Library library can be downloaded at this website.
    This package needs to be unzip and compiled (for instance) at /usr/include/petsc.

  Modify the Makefile.mk file so that the previous path point to the right libraries:
  

  MPI_DIR   = /usr/include/mpi
  EIGEN_DIR = /usr/include/eigen
  PETSC_DIR = /usr/lib/petsc

  Also, make sure libraries such as: -lscalapack-openmpi -lblacs-openmpi -llapack -lblas and -lparmetis -lmetis are also installed. Refer to the Ubuntu Operating System example to see all the required dependencies. Finally, inside SVL/02-Run_Process directory, write in terminal:

  make -s DEBUG=False

Note

This installation process assumes the user is able to compile/build PETSc and MUMPS from the source code provided.
In MacOS (Unix) a similar described for the Ubuntu Operating System process can be followed using brew.