Installation instructions for Mac users
Mac users can follow the same instructions as linux users, except for the installation of the external dependencies (requirements):
Requirements
Minimum
C++ and Fortran compilers
Mercury uses C++14 features and thus requires an up-to-date compiler.
We suggest you use clang, or gcc 5.0 or higher.To compile our Fortran files, we recommend gfortran.
cmake version 2.8.14 or higher (http://cmake.org/cmake)
python version 3
git version 2.2 or higher
Optional
doxygen and graphviz (dot) used to create a local copy of the documentation
graphviz (https://graphviz.gitlab.io/download/)
doxygen (http://www.doxygen.nl/download.html)
X11 used to visualise your simulation results with xballs
Paraview used to visualise your simulation results, rendered in 3D
OpenMPI libraries for parallel computing
Installation steps
Step 1: Install external dependencies
To install an up-to-date c++ compiler we recommend you to install either XCode or the Command Line Tools. XCode provides you with a nice IDE, but is ~10GB in size, whereas the Command line tools is a much lighter package. To install the Command Line Tools, open a terminal and type
xcode-select --install
For more details, see https://osxdaily.com/2014/02/12/install-command-line-tools-mac-os-x/.
To install the other dependencies, we recommend you to use a package manager such as brew. Once you have installed brew, you can install the other packages as follows:
brew install --cask cmake
brew install git
brew install python
brew install gcc
Notes: the "--cask" version of cmake adds the graphical user interface; gcc is installed to get gfortran, you can install any other fortran compiler.
Similarly, you also can install the optional dependencies, like doxygen:
brew install --cask xquartz
brew install doxygen
brew install graphviz
Step 2: Install MercuryDPM
Follow the installation instructions for linux to install MercuryDPM, starting with step 2.
Note: Starting with the release of Sonoma, 16 self-tests currently fail on MacOS with M1 processors. We are working on resolving this issue. For now, please ignore the failing selftests.