noloadj 1.2.4

solving constrained optimization problem for the design of engineering systems

NoLOAD_Jax: Non Linear Optimization by Automatic Differentiation using Jax

We are happy that you will use or develop the NoLOAD_Jax. It is an Open Source project located on GitLab at https://gricad-gitlab.univ-grenoble-alpes.fr/design_optimization/NoLoad_v2 It aims at solving constrained optimization problem for the design of engineering systems

Project Presentation

NoLOAD_Jax: Please have a look to NoLOAD presentation : https://noload_jax.readthedocs.io/en/latest/

NoLOAD_Jax Community

Please use the git issues system to report an error: https://gricad-gitlab.univ-grenoble-alpes.fr/design_optimization/NoLoad_v2 Otherwise you can also contact the developer team using the following email adress: [e-mail]

Installation Help

You can install the library as a user or as a developer. Please follow the corresponding installation steps below.

Prerequisite

Please install Python 3.8 or later https://www.python.org/downloads/ You must run NoLOAD_Jax on Ubuntu : on Jupyter Notebook on a computer using Ubuntu, or as explained later, using WSL (Windows Subsystem for Linux) if your computer works on Windows. As it uses the JAX library, NoLOAD_Jax can run on CPU (Central Processor Unit) or GPU (Graphics Processor Unit), where GPU offers better performances than CPU. With WSL only CPU can be used. To use GPU you may run NoLOAD on Ubuntu. If you want to use GPU, you need to install CUDA and CuDNN on your computer then tape on Pycharm terminal (where 0.3.XX is your JAX version):

pip install --upgrade pip
pip install --upgrade jax jaxlib==0.3.XX+cuda111 -f https://storage.googleapis.com/jax-releases/jax_releases.html

If you use GPU, you need to put these lines at the beginning of your "optimization" file to avoid memory issues :

import os
os.environ['XLA_PYTHON_CLIENT_PREALLOCATE']='false'
os.environ['XLA_PYTHON_CLIENT_MEM_FRACTION']='0.50'

To install IPOPT algorithm, please install an Anaconda environment and run this command on a terminal :

conda install -c conda-forge cyipopt

How to configure a virtual environment on WSL running with a Ubuntu distribution :

At first, activate the WSL on you computer and install on it a Ubuntu distribution by following the 6 first steps of the link : https://docs.microsoft.com/en-us/windows/wsl/install-win10

Then open WSL.exe using the Windows search bar and tape :

cd ~
sudo apt update
sudo apt install python3-pip
sudo pip3 install virtualenv
python3 -m virtualenv pythonenv

You can replace "pythonenv" by "venv" or another name of your choice. Then write :

source pythonenv/bin/activate

You must see a "(pythonenv)" written at the beginning of the command line. Now close WSL and open Pycharm in a new project. Go to File/Settings/Project:name_of_your_project/Project Interpreter and add a new interpreter. Choose WSL option then change the Python interpreter path by /home/your_username/pythonenv/bin/python3. (your_username is the login you chose for WSL) Close the Settings window. Click on Terminal section at the bottom, and tape on the commande line :

source /home/your_username/pythonenv/bin/activate

If you see a "(pythonenv)" written at the beginning of the command line, the configuration is completed and you can run your code now !

Installation as a user

Please install NoLOAD_Jax with pip using the command prompt.

If you are admin on Windows or working on a virtual environment

pip install noloadj

If you want a local installation or you are not admin

pip install --user noloadj

If you are admin on Linux:

sudo pip install noloadj

Launch the examples to understand how the NoLOAD_Jax works:

python noload/01-UnconstrainedMonoObjective.py
python noload/02-ConstrainedMonoObjective.py
python noload/03-ConstrainedMultiObjective.py
python noload/04-ConstrainedMonoObjective2.py

Enjoy your time using NoLOAD_Jax !

Library Installation Requirements

Matplotlib >= 3.0 Scipy >= 1.2 Jax >= 0.3.25 Jaxlib >= 0.3.25 Pandas >= 1.3.5

Main Authors:

B. DELINCHANT, L. GERBAUD, F. WURTZ, L. AGOBERT

Partners:

Vesta-System: http://vesta-system.fr/

Acknowledgments:

Licence

This code is under the Apache License, Version 2.0