gemseo_umdo

gemseo_umdo.formulations

• factory
• formulation
• sampling
• sequential_sampling
• taylor_polynomial

Formulations for multidisciplinary design problems under uncertainty.

An [MDOFormulation][gemseo.core.formulation.MDOFormulation] defines an [OptimizationProblem][gemseo.algos.opt_problem.OptimizationProblem] from one or several [MDODiscipline][gemseo.core.discipline.MDODiscipline]s, a [DesignSpace][gemseo.algos.design_space.DesignSpace], an objective and constraints. The objective can be either minimized (default) or maximized.

In the context of deterministic MDO, the [OptimizationProblem][gemseo.algos.opt_problem.OptimizationProblem] is handled by a driver (see [DriverLibrary][gemseo.algos.driver_library.DriverLibrary]), typically an optimizer (see [OptimizationLibrary][gemseo.algos.opt.optimization_library.OptimizationLibrary]), or a design of experiments (DOE, see [DOELibrary][gemseo.algos.doe.doe_library.DOELibrary]).

In the frame of robust MDO, the [UMDOFormulation][gemseo_umdo.formulations.formulation.UMDOFormulation] uses a [MDOFormulation][gemseo.core.formulation.MDOFormulation] with a [ParameterSpace][gemseo.algos.parameter_space.ParameterSpace] defining the uncertain variables and executes the corresponding [OptimizationProblem][gemseo.algos.opt_problem.OptimizationProblem] with a particular DOE. Then, it post-processed the associated [Database][gemseo.algos.database.Database] to estimate the statistics applied to the objective and constraints.

The most common [UMDOFormulation][gemseo_umdo.formulations.formulation.UMDOFormulation] is [Sampling][gemseo_umdo.formulations.sampling.Sampling], consisting in estimating the statistics with (quasi) Monte Carlo techniques.