gemseo-umdo

Overview

gemseo-umdo is a plugin of the library GEMSEO, dedicated to multidisciplinary optimization (MDO) under uncertainty.

MDO under uncertainty

The main goal of gemseo-umdo is to extend GEMSEO to MDO under uncertainty.

Given a collection of disciplines, we are interested in solving a problem like

$$ \begin{align} &\underset{x\in\mathcal{X}}{\operatorname{minimize}}& & \mathbb{E}[f(x,U)]+\kappa\times\mathbb{S}[f(x,U)] \ &\operatorname{subject;to} & &\mathbb{P}[g(x,U)\geq 0] \leq \varepsilon \end{align} $$

by selecting an MDO formulation to handle the multidisciplinary coupling and an estimation technique to approximate the statistics.

Statistics

gemseo-umdo also proposes advanced techniques for uncertainty quantification and management (UQ&M). In presence of multilevel simulators, multilevel Monte Carlo (MLMC) sampling can reduce the variance of the statistics estimators. Another variance reduction technique consists of using the outputs of surrogate models as control variates, even moderately correlated with the original models.

Visualization

A third facet of gemseo-umdo is the visualization toolbox to display the propagation of the uncertainties through a multidisciplinary system as well as the interaction between the uncertain input variables.

Installation

Install the latest version with pip install gemseo-umdo.

See pip for more information.

Bugs and questions

Please use the gitlab issue tracker to submit bugs or questions.

Contributing

See the contributing section of GEMSEO.

Contributors

• Antoine Dechaume
• Matthias De Lozzo