🤖🌊🚀 AI-Driven Design and Optimization (ADOPT)

About

Python package for AI-driven design and optimization of scientific simulators building on

• 📈 Ray for parallelization, auto-scaling, and auto-tuning
• 🤖 Vizier Service to keep track of the optimization configurations

Belied by a metadata system guaranteeing the physical correctness of configurations, and employing an SMT-solver to solve for allowed configurations of solvers, and optimization algorithms.

Installation

To get the full repository you have to either recursively clone the repository with its submodules

git clone --recursive https://github.com/adopt-opt/adopt-py

or if you already have an adopt-py download, initialize the submodules

git submodule update --init --recursive

To then set up the conda environment we use the environment.yml:

conda env create -f environment.yml

The adopt-py package itself can then be installed with:

pip install .

For package development you also need to install requirements-dev.txt with pip install -r requirements-dev.txt.

Contributing

The workflow to contribute code to the package is described in minute detail in Contributing.

Packaged Components

Solvers

• jaxfluids: compressible fluid dynamics in JAX

Design Optimization Algorithms