Initial Commit

README.md

@@ -1,10 +1,24 @@
 # Predicting Turbulent Fluid Flows with Generative Deep Learning
-Multi-fidelity Generative Deep Learning Turbulent Flows [[JCP]()][[ArXiv](https://arxiv.org/abs/2006.04731)]
+Multi-fidelity Generative Deep Learning Turbulent Flows [[FoDS]()][[ArXiv](https://arxiv.org/abs/2006.04731)]
 
 [Nicholas Geneva](http:https://nicholasgeneva.com/), [Nicholas Zabaras](https://cics.nd.edu)
 
 ---
+[![Documentation Status]()]() ![liscense](https://img.shields.io/github/license/zabaras/deep-turbulence)
 
-This paper is currently in the review process. The code is also being prepared for release in the near future.
+A novel multi-fidelity deep generative model is introduced for the surrogate modeling of high-fidelity turbulent flow fields given the solution of a computationally inexpensive but inaccurate low-fidelity solver.
 
-Star this repository for updates. Thank you for your patience.
+- Getting Started
+- Documentation 
+- Data Repository
+
+## Citation
+Cite us with:
+```latex
+ @article{geneva2020multi,
+ title={Multi-fidelity Generative Deep Learning Turbulent Flows},
+ author={Geneva, Nicholas and Zabaras, Nicholas},
+ journal={arXiv preprint arXiv:2006.04731},
+ year={2020}
+ }
+```

docs/Makefile

@@ -0,0 +1,20 @@
+# Minimal makefile for Sphinx documentation
+#
+
+# You can set these variables from the command line, and also
+# from the environment for the first two.
+SPHINXOPTS ?=
+SPHINXBUILD ?= sphinx-build
+SOURCEDIR = source
+BUILDDIR = build
+
+# Put it first so that "make" without argument is like "make help".
+help:
+ @$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
+
+.PHONY: help Makefile
+
+# Catch-all target: route all unknown targets to Sphinx using the new
+# "make mode" option. $(O) is meant as a shortcut for $(SPHINXOPTS).
+%: Makefile
+ @$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)

docs/make.bat

@@ -0,0 +1,35 @@
+@ECHO OFF
+
+pushd %~dp0
+
+REM Command file for Sphinx documentation
+
+if "%SPHINXBUILD%" == "" (
+ set SPHINXBUILD=sphinx-build
+)
+set SOURCEDIR=source
+set BUILDDIR=build
+
+if "%1" == "" goto help
+
+%SPHINXBUILD% >NUL 2>NUL
+if errorlevel 9009 (
+ echo.
+ echo.The 'sphinx-build' command was not found. Make sure you have Sphinx
+ echo.installed, then set the SPHINXBUILD environment variable to point
+ echo.to the full path of the 'sphinx-build' executable. Alternatively you
+ echo.may add the Sphinx directory to PATH.
+ echo.
+ echo.If you don't have Sphinx installed, grab it from
+ echo.http:https://sphinx-doc.org/
+ exit /b 1
+)
+
+%SPHINXBUILD% -M %1 %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O%
+goto end
+
+:help
+%SPHINXBUILD% -M help %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O%
+
+:end
+popd

docs/source/conf.py

@@ -0,0 +1,73 @@
+# Configuration file for the Sphinx documentation builder.
+#
+# To auto-generate rst files for the program use the command:
+# sphinx-apidoc -f -o ./source/tmglow/ ../tmglow/
+# Use "make clean" and "make html" to build the docs
+
+import sphinx_rtd_theme
+
+# -- Path setup --------------------------------------------------------------
+
+# If extensions (or modules to document with autodoc) are in another directory,
+# add these directories to sys.path here. If the directory is relative to the
+# documentation root, use os.path.abspath to make it absolute, like shown here.
+#
+import os
+import sys
+sys.path.insert(0, os.path.abspath('.'))
+sys.path.insert(0, os.path.abspath('../../tmglow'))
+
+from sphinx.ext.autodoc import between
+
+def setup(app):
+ # Register a sphinx.ext.autodoc.between listener to ignore everything
+ # between lines that contain the word ======
+ app.connect('autodoc-process-docstring', between('^.*=====.*$', exclude=True))
+ return app
+
+# -- Project information -----------------------------------------------------
+
+project = 'Multi-fidelity Generative Deep Learning Turbulent Flows'
+copyright = '2020, Nicholas Geneva'
+author = 'Nicholas Geneva'
+
+# The full version, including alpha/beta/rc tags
+release = '0.0.1'
+
+
+# -- General configuration ---------------------------------------------------
+
+# Add any Sphinx extension module names here, as strings. They can be
+# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
+# ones.
+extensions = [
+ "sphinx.ext.autodoc",
+ "sphinx.ext.doctest",
+ 'sphinx.ext.todo',
+ "sphinx_rtd_theme",
+ "sphinx.ext.mathjax",
+ 'sphinx.ext.ifconfig'
+]
+
+
+# Add any paths that contain templates here, relative to this directory.
+templates_path = ['_templates']
+
+# List of patterns, relative to source directory, that match files and
+# directories to ignore when looking for source files.
+# This pattern also affects html_static_path and html_extra_path.
+exclude_patterns = []
+
+mathjax_path = "https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"
+
+# -- Options for HTML output -------------------------------------------------
+
+# The theme to use for HTML and HTML Help pages. See the documentation for
+# a list of builtin themes.
+#
+html_theme = 'sphinx_rtd_theme'
+
+# Add any paths that contain custom static files (such as style sheets) here,
+# relative to this directory. They are copied after the builtin static files,
+# so a file named "default.css" will overwrite the builtin "default.css".
+html_static_path = ['_static']

docs/source/faq.rst

@@ -0,0 +1,46 @@
+F.A.Q.
+=======
+Here are a few questions that may perhaps come up.
+
+- **Where can I download the training/testing data?**
+
+ Check the :ref:`getting_started` section for those details.
+
+- **What Python packages do I need?**
+
+ Please see the requirements.txt or readme on Github.
+
+- **Where in the code of the actual TM-Glow model?**
+
+ The class you want to start at is :class:`nn.tmGlow.TMGlow`.
+
+- **Why use an INN instead of a VAE or GANs?**
+
+ VAEs have long suffered a lack of ability to yield crisp, complex images.
+ GANs can be extremely unstable and were tested at the early stages of this work
+ without success.
+ Additionally, the INN has very nice probabilistic properties which is good for UQ.
+
+- **My model is unstable during training!?!? Help!**
+
+ Restart training at a checkpoint before your model diverged with :code:`--epoch-start #` parameter.
+ Consider increasing gradient clipping, lower learning rate, increasing weight decay, etc. There are a lot
+ of options to play with. Which Glow is much more stable than GANs it can still be unstable at points.
+
+- **Why are some of the sections of the program not fully documented?**
+
+ Documenting takes time and energy. I'm working on slowly improving the docs, but it was my main focus to get
+ the model and its components documented so people can understand what is going on under the hood.
+
+- **How should I cite this?**
+
+ Cite the paper!
+
+ .. code-block:: text
+
+ @article{geneva2020multi,
+ title={Multi-fidelity Generative Deep Learning Turbulent Flows},
+ author={Geneva, Nicholas and Zabaras, Nicholas},
+ journal={arXiv preprint arXiv:2006.04731},
+ year={2020}
+ }

docs/source/help.rst

@@ -0,0 +1,3 @@
+Contact
+=======
+Have a question, issue or found a bug? Best way to get into contact is through an issue on our github page!

docs/source/index.rst

@@ -0,0 +1,42 @@
+
+Multi-fidelity Generative Deep Learning Turbulent Flows
+===================================================================================
+Nicholas Geneva, Nicholas Zabaras
+
+
+Documentation transient multi-fidelity Glow (TM-Glow) used for predicting high-fidelity turbulence
+given a low-fidelity simulation.
+Please see the paper for additional information regarding the model used and the underlying
+theory `[JCP] <https://arxiv.org/abs/2006.04731>`_ `[Arxivs] <https://arxiv.org/abs/2006.04731>`_.
+
+User Guide
+-------------
+.. toctree::
+ :maxdepth: 2
+
+ start
+ faq
+
+API Reference
+-------------
+
+If you are looking for information on a specific function, class or
+method, this part of the documentation is for you.
+
+.. toctree::
+ :maxdepth: 2
+
+ tmglow/main
+ tmglow/args
+ tmglow/nn
+ tmglow/pc
+ tmglow/utils
+
+Additional Information
+-------------
+.. toctree::
+ :maxdepth: 2
+
+ resources
+ help
+ liscense

docs/source/liscense.rst

@@ -0,0 +1,24 @@
+License
+=======
+
+MIT License
+
+Copyright (c) 2020 Scientific Computing and Artificial Intelligence 
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

docs/source/resources.rst

@@ -0,0 +1,11 @@
+Resources
+=======
+For additional information on the use of invertible neural networks, please refer to the following select papers:
+
+- D. Kingma, P. Dhariwal, Glow: Generative flow with invertible 1x1 convolutions, in: Advances in Neural Information Processing Systems, 2018, pp. 10215–10224. `[Link] <http:https://papers.nips.cc/paper/8224-glow-generative-flow-with-invertible-1x1-con>`_
+
+- L. Dinh, J. Sohl-Dickstein, S. Bengio, Density estimation using real nvp, arXiv preprint arXiv:1605.08803 `[Link] <https://arxiv.org/abs/1605.08803>`_
+
+- L. Dinh, D. Krueger, Y. Bengio, Nice: Non-linear independent components estimation, arXiv preprint arXiv:1410.8516. `[Link] <https://arxiv.org/abs/1410.8516>`_
+
+- Y. Zhu, N. Zabaras, P.-S. Koutsourelakis, P. Perdikaris, Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data, Journal of Computational Physics 394 (2019) 56 – 81. `[Link] <https://doi.org/10.1016/j.jcp.2019.05.024>`_

docs/source/start.rst

@@ -0,0 +1,56 @@
+.. _getting_started:
+
+Getting Started
+===============
+
+1. Clone the repository
+-----------------------
+Start with cloning the repository to your local machine using:
+ ``git clone [email protected]:zabaras/deep-turbulence.git``
+
+2. Downloading Data
+-------------------
+To download both the training and testing data for both of the numerical examples, visit the following Zenodo repository.
+`https://doi.org/10.5281/zenodo.4298896 <https://doi.org/10.5281/zenodo.4298896>`_
+
+Extract the data using ``tar –xvzf data-set-name.tar.gz``, and move the training and testing data the desired directory.
+
+The default directories in the config are:
+
+- ``/deep-turbulence/cylinder-training/``
+- ``/deep-turbulence/cylinder-testing/``
+- ``/deep-turbulence/step-training/``
+- ``/deep-turbulence/step-testing/``
+
+But custom paths can be easily adjusted in the configuration file.
+
+3. Set-up Conda Python Environment
+----------------------------------
+For your convenience, a requirements.txt file is provided in ``/tmglow/`` which will allow the easy creation of a conda environment that contains
+the required packages used. 
+
+``conda create --name <env_name> --file requirements.txt``
+
+4. Start Training
+-----------------
+To start training the model simply run :doc:`tmglow/main`. 
+There are many customizable parameters you can change in :doc:`tmglow/args` which can be view with ``python main.py --help``. 
+Both numerical examples have pre-coded
+configurations.
+
+For example:
+
+- Train the model for the backward-step example: ``python main.py --exp-type backward-step``
+- Train the model for the backward-step example: ``python main.py --exp-type cylinder-array``
+- Use a custom training directory: ``python main.py --training_data_dir <custom directory>``
+- Change number of epochs: ``python main.py --epochs <# epochs>``
+
+.. warning::
+ Memory constraints on the GPU will likely be of concern. Parallel training is supported but only on a single 
+ node (not multiple CPU) which can be controlled through ``--n_gpu`` and ``--parallel`` options. 
+ Support is not provided for debugging GPU memory or parallel issues.
+
+4. Running Pre-trained Models
+-----------------------------
+Alternatively, you can skip the training and simply run a pre-trained models.
+These can be found in the ``example`` folder where there are several scripts that demonstrate how to load and test a model.

docs/source/tmglow/args.rst

@@ -0,0 +1,7 @@
+args module
+===========
+
+.. automodule:: args
+ :members:
+ :undoc-members:
+ :show-inheritance:

docs/source/tmglow/main.rst

@@ -0,0 +1,7 @@
+main module
+===========
+
+.. automodule:: main
+ :members:
+ :undoc-members:
+ :show-inheritance:

docs/source/tmglow/modules.rst

@@ -0,0 +1,11 @@
+tmglow
+======
+
+.. toctree::
+ :maxdepth: 4
+
+ args
+ main
+ nn
+ pc
+ utils