Flume

Flume is an open-source framework for multidisciplinary analysis and adjoint evaluations based on directed acyclic graphs. It is intended to be a lightweight framework that affords users with a degree of flexibility in implementing their own analysis and optimization procedures while providing a common underlying structure. Flume also has a set of interfaces between some common optimizers, which allow the user to easily setup an optimization problem that corresponds to their specific needs.

Getting Started

For new users, it is recommended that you read the Flume overview notebook, which outlines in detail the critical aspects of the framework in context of an example. This discusses the nomenclature that is used within the framework and illustrates how to construct State, Analysis, and System objects for use with a Flume interface. Also, for a review of other useful methods that are included for Analysis classes, check out the methods demonstration notebook. The examples gallery also includes several different types of problems, which showcase the construction of many types of Analysis classes, the assembly of a System, and how to interface with an optimizer interface.

Installation

To install Flume, the only necessary dependencies are a working version of Python that is at least version 3.10 and Graphviz. It is generally encouraged that you use a virtual environment. If you do not have a virtual environment already, navigate to the directory where you want to use Flume and execute the following.

python -m venv venv
source venv/bin/activate

Then, with the virtual environment activated, simply install the package from the GitHub repository with

pip install git+https://github.com/smdogroup/flume.git

This will install the base classes, optimizer interfaces, and utilities into a Python package named “flume”. For those who want to make changes to the code or access the unit tests and examples, you can also clone the repository and then install in editable mode.

cd {your_chosen_git_directory}
git clone https://github.com/smdogroup/flume.git
cd flume
python -m venv venv
source venv/bin/activate
pip install -e .

To install Graphviz, check out the installation instructions located here depending on your machine type.

After following either of these processes, you should be able to access the base classes and get started with developing your own scripts within the framework! It should be noted that ParOpt is a separate package that is not included with Flume during the build process. For those who want to utilize this interface, we refer you to the ParOpt installation instructions.