Installation

Migflow can be installed as a python package or compiled from source. In addition to migflow itself, the gmsh mesh generator is required and optionally petsc4py.

Migflow

Python package

The easiest way to install migflow is to use the package from pypi using pip. If you do not have pip installed (check the output of pip –version), follow the official instruction to get it. Once pip is installed, simply type

python3 -m pip install migflow

Or optionally, a virtual environment can be used

python3 -m venv migflow-env
source migflow-env/bin/activate
python3 -m pip install migflow

GMSH

Meshes representing the computational domain, on which Navier-Stokes equations are discretised, are created from .geo files using GMSH.

If you use a package manager, it is recommended to install GMSH from your package manager (e.g. apt install gmsh on ubuntu).

Otherwise, you can download it from GMSH website or install it with pip (python3 -m pip install gmsh). Either way, check that the gmsh executable is in your PATH variable.

ParaView

ParaView is an open-source, multi-platform data analysis and visualization application.

If you use a package manager, it is recommanded to install ParaView from your package manager (e.g. apt install paraview on ubuntu).

Otherwise, you can download it from ParaView website.

Plugins

In order to install MigFlow Plugins, you need to open ParaView. Access “Tools → Manage Plugins…”. On the opened window, click on “Load New Plugin” and select “migflow/paraview_utils/migflow_plugins.py”. You will need to tick the “Auto-load” box.

Now, you are able to open the migflow extensions (.migf, .migs). If you load a solid problem (.migs), the solid reader will be called. To have the default visualization, click on the “Set Default” box into the Properties menu.

Linear Solver

MigFlow uses a linear solver to solve Navier-Stokes. The available linear solvers are :

• Pardiso

• Scipy

• PETSc4py

• PETSc

• MUMPS

The recommended solver is Pardiso (MKL).

Pardiso Installation

Pardiso is a parallel direct solver designed for large sparse matrix (Pardiso webpage).

It can be install using pip : python3 -m pip install mkl

Scipy Installation

Scipy provides linear solver but its efficiency is limited on large matrices. If you consider large meshes, other linear solver should be used.

It can be install using a package manager : apt install python3-scipy or using pip : python3 -m pip install scipy.

PETSc4py Installation

The installation of PETSc4py is optional, if it is not available, migflow will use pardiso instead.

Again if you use a package manager and if a PETSc4py package is available, use it. (ubuntu : apt install python3-petsc4py).

Otherwise, follow the instructions bellow to compile it from sources.

1. For other OS, go on PETSc webpage and clone the repository on your computer:

   git clone -b release https://gitlab.com/petsc/petsc.git petsc
   

2. Compiling options for PETSc:

   cd petsc
   
   ./configure --with-debugging=0 --with-shared-libraries=1 COPTFLAGS="-O2" FOPTFLAGS="-O2" CXXOPTFLAGS="-O2" --with-mpi=0 --with-petsc4py=yes --with-fortran=0 --download-f2cblaslapack=1
   

3. Compile PETSc:

   make -j4 install
   

4. To use PETSc4py, you have to state the library location by writing the following into your .bashrc

   export PYTHONPATH=/petsc/arch-linux-c-opt/lib
   

MUMPS Installation

MUMPS is a multifrontal massively parallel solver with high efficiency for large systems of sparse linear equations (MUMPS webpage).

It can be install when you’re building your build directory:

mkdir build

cmake -DBUILD_MUMPS=ON ..

make -j4

Compilation from sources

If you plan to develop migflow, you’ll have to compile it from source (instead of installing the python package). if your operating system is Windows 10 version 2004 or higher, use the Windows Subsystem for Linux.

1. Go on the gitlab page dedicated to the project to upload your ssh or https link

2. Open your terminal and type the following commands:

   cd ~
   
   git clone https://git.immc.ucl.ac.be/fluidparticles/migflow.git
   
   cd migflow
   
   mkdir build
   
   cd build
   
   cmake ..
   
   make -j8
   

3. State the location of the python modules by writing the following line in your .bashrc (For MacOS you possibly need to create this hidden file in your HOME directory):

   export PYTHONPATH=$HOME/migflow/build:$PYTHONPATH