Documentation

Because github pages are publicly accessible, I moved the documentation to the "doc" branch. To download the documentation either

1. Download the zip file from the "doc" branch

2. clone the repo with

git clone -b doc http:// https://personal_access_token@github.com/lukaskiwitz/thesis

To clone the repo from the cli you will need to set up a personal acces token in your github settings. Then you can download any update with

git pull

3. Alternatively, you can build the documentation yourself, provided that you have a properly configured conda environment. Move to the docsrc folder and run

make html

When done open index.html in your web browser to access the title page.

Requirements

1. All python dependencies (provided as an anaconda yml-file)
2. correctly set LD_LIBRARY_PATH environment variable
3. installed thesisproject-package (provided as wheel file for pip)

Installation Instructions (with anaconda)

1. Install anaconda: https://www.anaconda.com/products/individual

2. Run

conda env create -f fenics.yml -n "enviroment_name"

to create a new anaconda environment with the necessary packages. This could take a moment.

3.Run

conda activate "enviroment_name"

to activate the newly created environment.

4.Install the package using pip with

pip install wheel-file

if you got the wheel file or

pip install -e "repository_dir"

if you cloned this repository.

5.Run

conda env config vars set LD_LIBRARY_PATH=~/anaconda3/envs/enviroment_name/lib

and if you want to use paraview rendering

conda env config vars set PARAVIEW_PATH=~path_to_paraview_dir

to set up necessary environment variables.

6.Reactivate the conda environment to apply changes

conda activate enviroment_name

Usage

1.Create a copy of one of the folders in "example_scripts/" in a location of your choice.

2.Open the parameters.py file and edit the path variable at the bottom to point to the desired top level directoy for your simulation results. Note: If your are on the itb computers this will work well without a change.

3.Run

For simple models run

python run.py; python post_process.py number_of_threads;

For bigger scans a run_all.py and list_post_process.py is available. We recommend a HPC to run

python run_all.py; python list_post_process.py; python create_df.py; python combine_dfs.py

The first time any model is run it will need to

• create a mesh
• mark the boundary parts
• run the FFC compiler

this may take some time. For large meshes this can be considerable (1h or so), but subsequent runs should be much faster.

Paraview Integration

Only works for simulation data generated with the current version

1. Download an omesa build of paraview (necessary to run without X-server)
2. Set the PARAVIEW_PATH env. variable to your paraview directory. For instance in you conda env. or in the post_process.py
3. pass the keyword argument "render_paraview=True" to run_post_process()
4. various settings can be passed. "example_min" contains a commented example

The cell type markers are floats and will be set according to "SimContainer.marker_lookup". SimContainer now has the "markers" attribute, which lists entity properties (parameters or python attributes) for which to export a MeshFunction. Assigment of non numerical values is controlled by marker_lookup.