FEALPy:Finite Element Analysis Library in Python

While beginning with the finite element algorithm, FEALPy's sights are set on exploring vast horizons.

We hope FEALPy will be an open-source library for intelligent CAE simulation algorithms, integrating CAE fundamentals with AI to support advanced algorithm research and the cultivation of versatile talent.

We also hope FEALPy can accelerate the creation and testing of next-gen intelligent CAE apps, paving the way for advanced algorithms in industrial applications.

So FEALPy's development goal is to become the next generation intelligent CAE simulation computing engine.

The word "FEAL" is an archaic or poetic term in English, meaning faithful or loyal. Though not commonly used in modern English, it carries strong connotations of unwavering dedication and reliability.

The name "FEALPy" embodies this essence of loyalty and faithfulness. It signifies the software's commitment to being a dependable and trustworthy tool in the field of intelligent CAE simulation. Just as "FEAL" suggests steadfastness, FEALPy aims to provide consistent, reliable support for researchers, engineers, and developers in their pursuit of innovative solutions and advancements in CAE simulations. The name reflects the software's mission to be a loyal companion in the journey toward groundbreaking discoveries and industrial applications.

Installation

Miniconda

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh
~/miniconda3/bin/conda init bash

conda create -n gpufealpy310 python=3.10
conda activate gpufealpy310
conda install numpy=2.0.1 -c conda-forge #2.0.1
conda install ipython notebook -c conda-forge
conda install jaxlib=*=*cuda* jax cuda-nvcc -c conda-forge -c nvidia # 0.4.31
conda install cupy -c conda-forge  -c nvidia
conda install pytorch=2.3.1 -c conda-forge -c nvidia

From Source (Recommanded)

First, clone the FEALPy repository from GitHub

git clone https://github.com/weihuayi/fealpy.git

If you can't acess GitHub, you can clone it from Gitee

git clone https://gitee.com/whymath/fealpy

It is recommended to create a virtual environment to manage dependencies:

python -m venv fealpy_env
source fealpy_env/bin/activate  # On Windows, use `fealpy_env\Scripts\activate`

Then change directory to the cloned repository and install FEALPy in editable(-e) mode:

cd fealpy
pip install -e .

If you want to install optional dependencies, such as pypardiso, pyamg, meshpy and so on, you can do so by specifying the [optional] extra:

pip install -e .[optional]

To install both development and optional dependencies, use:

pip install -e .[dev,optional]

To verify that FEALPy is installed correctly, you can run the following command:

python -c "import fealpy; print(fealpy.__version__)"

To update your FEALPy installation to the latest version from the source repository, navigate to the FEALPy directory and pull the latest changes:

cd fealpy
git pull origin main

To uninstall FEALPy, just run the following command:

pip uninstall fealpy

Development

For FEALPy developers, the first step is to create a fork of the https://github.com/weihuayi/fealpy repository in your own Github account.

Clone the FEALPy repository under your own account to the local repository:

# replace<user name>with your own GitHub username
git clone git@github.com:<user name>/fealpy.git

Note that the following operations need to be operated in the fealpy folder.

Set up the upstream repository:

git remote add upstream git@github.com:weihuayi/fealpy.git

Before local development, need to pull the latest version from the upstream repository and merge it into the local repository:

git fetch upstream
git merge upstream/master

After local development, push the modifications to your own remote repository:

git add modified_files_name
git commit -m"Explanation on modifications"
git push

Finally, in your own Github remote repository, open a pull request to the upstream repository and wait for the modifications to be merged.

Warning

The sparse pattern of the matrix A generated by FEALPy may not be the same as the theoretical pattern, since there exists nonzero values that are close to machine precision due to rounding. If you care about the sparse pattern of the matrix, you can use the following commands to eliminate them

eps = 10**(-15)
A.data[ np.abs(A.data) < eps ] = 0
A.eliminate_zeros()

Docker

To be added.

Reference and Acknowledgement

We thank Dr. Long Chen for the guidance and compiling a systematic documentation for programming finite element methods.

Citation

Please cite fealpy if you use it in your paper

H. Wei and Y. Huang, FEALPy: Finite Element Analysis Library in Python, https://github.com/weihuayi/fealpy, Xiangtan University, 2017-2024.

@misc{fealpy,
    title = {FEALPy: Finite Element Analysis Library in Python. https://github.com/weihuayi/fealpy},
    url = {https://github.com/weihuayi/fealpy},
    author = {Wei, Huayi and Huang, Yunqing},
    institution = {Xiangtan University},
    year = {Xiangtan University, 2017-2024},
}

weihuayi / fealpy

readme