HENKES_GAN

Code of the publication "Three-dimensional microstructure generation using generative adversarial neural networks in the context of continuum micromechanics" published in https://doi.org/10.1016/j.cma.2022.115497 by Alexander Henkes and Henning Wessels from TU Braunschweig.

Please cite the following paper:

@article{henkes2022three,
  title={Three-dimensional microstructure generation 
         using generative adversarial neural networks 
         in the context of continuum micromechanics},
  author={Henkes, Alexander and Wessels, Henning},
  journal={Computer Methods in Applied Mechanics and Engineering},
  volume={400},
  pages={115497},
  year={2022},
  publisher={Elsevier}
}

... and the code using the CITATION.cff file.

Requirements

The requirements can be found in

requirements.txt

and may be installed via pip:

pip install -r requirements.txt

Docker image

You can download a pre-built Docker image via:

docker pull ahenkes1/gan:1.0.0

If you want to build the Docker image, the official TensorFlow image is needed:

https://www.tensorflow.org/install/docker

Build via

docker build -f ./Dockerfile --pull -t ahenkes1/gan:1.0.0 .

Execute via

docker run --gpus all -it -v YOUR_LOCAL_OUTPUT_FOLDER:/home/docker_user/src/save_files/ --rm ahenkes1/gan:1.0.0 --help

where 'YOUR_LOCAL_OUTPUT_FOLDER' is an absolute path to a directory on your system. This will show the help.

Execute the code using standard parameters as

docker run --gpus all -it -v YOUR_LOCAL_OUTPUT_FOLDER:/home/docker_user/src/save_files --rm ahenkes1/gan:1.0.0 

Using XLA

The code may run using XLA (faster) using the following flag:

XLA_FLAGS=--xla_gpu_cuda_data_dir=/usr/local/cuda-11.2 python3 main.py --help

where the correct cuda path and version have to be used. The Docker image runs XLA natively.

GPU

The code uses mixed-precision. If your GPU has TensorCores, it will run much faster. Otherwise, a warning will be displayed. Nevertheless, the memory consumption is much lower in either case.

Tensorboard

The code logs several metrics during training, which can be accessed via Tensorboard. The logs can be found in the corresponding output folders.

https://www.tensorflow.org/tensorboard