femo is a general framework for using Finite Element in PDE-constrained Multidisciplinary Optimization problems. It relies on FEniCSx to provide solutions and partial derivatives of the PDE residuals, and uses CSDL as the umbrella for coupling and mathematical modeling of the whole problem. The code is still under active developement and we expect it to be available to the public for their research applications by 2023 Winter.
For modeling and simulation, you need to install FEniCSx
, CSDL
and the Python-based backend of CSDL
- python_csdl_backend; for optimization, you will also need ModOpt on top of them for the black-box optimizers.
It's recommended to use conda for installing the module and its dependencies.
- Create a conda environment for FEniCSx with a specific Python version (Python 3.9) that is compatible with all of the dependencies
(Python 3.9.7 also works if Python 3.9.10 is unavailable in your conda)
conda create -n fenicsx python=3.9.10
- Activate the conda enviroment
conda activate fenicsx
- Install FEniCSx
conda install -c conda-forge fenics-dolfinx=0.5.1
- Git clone and install CSDL, and python_csdl_backend by
pip
- Git clone and install femo by
pip
- (optional) Install SNOPT for optimization (licence required)
- (optional) Install ModOpt by
pip
and test withmodopt/modopt/external_packages/csdl/test_scaler.py
@misc{xiang2024,
author = "Xiang, Ru
and van Schie, Sebastiaan P.C.
and Scotzniovsky, Luca
and Yan, Jiayao
and Kamensky, David
and Hwang, John T.",
title = "Automating adjoint sensitivity analysis for multidisciplinary models involving partial differential equations",
howpublished = {Jul 2024, Preprint available at \url{https://doi.org/10.21203/rs.3.rs-4265983/v1}}
}
@misc{scotzniovsky2024,
author = "Scotzniovsky, Luca
and Xiang, Ru
and Cheng, Zeyu
and Rodriguez, Gabriel
and Kamensky, David
and Mi, Chris
and Hwang, John T.",
title = "Geometric Design of Electric Motors Using Adjoint-based Shape Optimization",
howpublished = {Feb 2024, Preprint available at \url{https://doi.org/10.21203/rs.3.rs-3941981/v1}}
}