Added docker (lululxvi#496)

docker/Dockerfile

@@ -0,0 +1,13 @@
+FROM horovod/horovod:master
+LABEL description="DeepXDE with Tensorflow PyTorch and GPU support"
+
+COPY requirements.txt /root
+WORKDIR /root
+RUN /usr/bin/python -m pip install --upgrade pip
+RUN pip3 install -r requirements.txt 
+RUN apt-get update && apt-get install -y --no-install-recommends imagemagick=8:6.9.7.4+dfsg-16ubuntu6.12 \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+EXPOSE 8888/tcp
+ENV SHELL /bin/bash
+ENTRYPOINT ["jupyter", "notebook", "--ip", "0.0.0.0", "--no-browser", "--allow-root"]

docker/requirements.txt

@@ -0,0 +1,11 @@
+ipython
+scipy
+sklearn
+matplotlib
+scikit-optimize
+tensorflow-probability==0.12.2
+jupyter
+jupyterlab
+notebook
+pandas
+deepxde

docs/user/installation.rst

@@ -32,6 +32,31 @@ Then, you can install DeepXDE itself.
     - `scikit-optimize <https://scikit-optimize.github.io>`_
     - `SciPy <https://www.scipy.org/>`_
 
+
+Docker
+------
+
+The `DeepXDE Docker images <https://hub.docker.com/r/pescapil/deepxde>`_ are already configured to run DeepXDE with the GPU support.
+
+You need first to install `NVIDIA Docker <https://github.com/NVIDIA/nvidia-docker>`_. 
+
+You can run a Jupyter Notebook environment with GPU-enabled stable DeepXDE using::
+
+    $ nvidia-docker run -v $(pwd):/root/shared -w "/root/shared" -p 8888:8888 pescapil/deepxde:latest
+
+Use ``nvidia-docker`` (resp. ``docker``) command to enable GPU (resp. CPU)-based DeepXDE. 
+
+The corresponding `Dockerfile <https://github.com/lululxvi/deepxde/tree/master/docker/Dockerfile>`_ is based on ``horovod/horovod`` and comes with TensorFlow 2.x and PyTorch. To build the image yourself, you can run::
+
+    $ git clone https://github.com/lululxvi/deepxde.git
+    $ cd deepxde/docker
+    $ docker build -f Dockerfile . -t mydeepxde
+
+and then run your own ``mydeepxde`` image via::
+
+$ nvidia-docker run -v $(pwd):/root/shared -w "/root/shared" -p 8888:8888 mydeepxde
+
+
 Working with different backends
 -------------------------------