Skip to content

Latest commit

 

History

History
152 lines (124 loc) · 6.52 KB

README.md

File metadata and controls

152 lines (124 loc) · 6.52 KB
Raw

Flows1D0D3D

CircleCI GitHub release GitHub license GitHub issues GitHub repo size

Table of contents

Introduction

Application which (currently) simulates bloodflows with coupled 1D, 0D and 3D models. Its future goal is to simulate and study breast cancer models.

Directory structure

  • apps: contains the demo applications (executibles are created inside <build path>/bin/macrocirculation)
  • assets: contains asset files for doxygen
  • cmake: contains cmake modules to locate petsc
  • docs: doxygen related files
  • external: keep external libraries
  • data: contains all the data needed for the simulations
    • 1d-boundary: Contains data for the calibrated RCR-models for different meshes.
    • 1d-coupling: Contains data coupling two geometries together.
    • 1d-input-pressures: Contains input pressures for other models to decoupled the simulations.
    • 1d-meshes: Contains the 1d networks.
    • 3d-meshes: Contains the 3d domains.
  • tools: contains utility scripts
    • mesh_creation: scripts for creating the 1D mesh
    • script/install: install scripts to facilitate setup
    • visualization: scripts for visualizing our output data
  • src: the source code of the library
  • tests: a few catch2 unittests for critical components
  • python: the python bindings
    • flows1d0d3d: the parts of our library written in python
    • examples: contains example applications in python calling our (already installed) library.

Installation

Dependencies

Core dependencies are:

  • cmake (3.12 or above)
  • petsc (3.13.3 or above)
    • see further below on how to build petsc
  • json (3.7.3)
    • directly pulled using CMake FetchContent module
    • for reading json file
  • cxxopts (2.2.1)
    • directly pulled using CMake FetchContent module
    • for reading command line options
  • Catch2 (2.13.1)
    • directly pulled using CMake FetchContent module
    • for ctest
  • gmm
    • included as external library in the code
    • provides framework to solve linear systems associated to the 1D networks
  • pybind11
    • included as a submodule in the code
    • provides simple python bindings to call our 1D and 0D solvers from python

Dependencies for running the examples:

  • python3
    • required to run the test python scripts
  • numpy
    • required to run the test python scripts
  • dolfin
    • required to run the 3d simulations for the microcirculation

Building the code

Assuming all dependencies are installed in global path (/usr/local/ etc), we build the code using

git clone --recurse-submodules https://github.com/CancerModeling/Flows1D0D3D.git

cd Flows1D0D3D && mkdir build && cd build

cmake   -DEnable_Documentation=ON \
        -DEnable_Tests=ON \
        -DCMAKE_BUILD_TYPE=Release \
        ..

make -j 4

cd tests

ctest --verbose

The --recurse-submodules parameter is required, if you want to use the python bindings and clones the pybind11 repository. If petsc is installed at a custom path one has to add -DPETSC_DIR="<petsc install path>" to the cmake parameters.

Recommendations for quick build

  1. Install most of the dependencies using apt-get:
sudo apt-get update 
  
sudo apt-get install -y build-essential ubuntu-dev-tools \
  wget curl lzip \
  cmake gfortran \
  libopenmpi-dev openmpi-bin \
  libboost-all-dev libvtk7-dev \
  liblapack-dev libblas-dev \
  doxygen doxygen-latex graphviz ghostscript \
  python3-pip 

# pyvista and pandas are not required, so they can be excluded
pip3 install numpy

  1. Build petsc. To learn how to build petsc on ubuntu 18.04 and 20.04, you can follow the dockerfile. Shell script could also be helpful.

  2. Use instructions in previous section to build Flows1D0D3D (provide right paths to libmesh and petsc).

Docker

For circle-ci testing, we use docker images prashjha/angio-base-bionic and prashjha/angio-base-focal of ubuntu 18.04 and 20.04 with petsc and libmesh installed. The associated dockerfiles can be found here.

In Packages, docker images of Flows1D0D3D is provided.

Installing the python bindings

For installation execute

PETSC_DIR=<REPLACE_WITH_PETSC_DIRECTORY> python3 -m pip install .

in the directory containing the setup.py file. Test if the installation was successful with

python3 -c "import flows1d0d3d"

For developers: Note that the providing the -e, --editable parameter to pip allows you to edit and test the python source code without having to reinstall the package.

Run unittests

The unittests can be run right after building the project with ctest

cmake <cmake-options-here> ..
make 
cd tests
ctest .

Developers