Follow us on Twitter
About
The ExaQUte project aims at constructing a framework to enable Uncertainty Quantification and Optimization Under Uncertainties in complex engineering problems, using computational simulations on Exascale systems.
ExaQUte General Objective
What?
Develop new
computational methods and
software tools.
Why?
To target Uncertainty Quantification and Optimization Under Uncertainties for Multiphysics and multiscale problems on geometrically complex domains.
How?
Taking advantage of
next‐generation
Exascale systems.
Goal 1
Develop a scheduling tool to extract parallelism in the MLMC algorithm across samples and levels.
Goal 2
Develop embedded solvers
for
multiphysics problems.
Goal 3
Develop parallel adaptive refinement methods for
embedded domains.
Goal 4
Develop space‐time methods
for the numerical simulation
of multiphysics problems.
Goal 5
Extend the MLMC
to use an adaptively refined space‐time mesh hierarchy.
Goal 6
Combine MLMC methods with gradient‐based optimization techniques based on adjoint problems.
Application
Shape optimization of civil engineering structures subjected to wind flow.
Partners
Workplan
Deriverables
WP 1 Embedded methods
WP 2 Mesh generation and adaptivity
D 1.1: Solvers “stub” implementation of the capabilities to be delivered (available under request)
D 1.2: First internal release of the solvers
D 1.3: First public Release of the solvers
D 1.4: Final public Release of the solvers
D 2.2: First release of the octree mesh-generation capabilities and of the parallel mesh adaptation kernel
D 2.3: Adjoint-Based error estimation routines
D 2.4: Second Release of the mesh generation/adaptation capabilities
D 2.5: Final Release of the mesh generation/adaptation capabilities
WP 3 Space-time parallelization
WP 4 Dynamic scheduling for MLMC
D 3.1: Report on nonlinear domain decomisition preconditioners and release of the solvers
D 3.2: Report on parallel in time methods and release of the solvers
D 3.3: Release of a 4D embedded mesh generation library
D 3.4: Report on adjoint-based space time error estimators
D 3.5: Report on the calibration of parallel space time algorithms for turbulent flows
D 4.2: Profiling report of the partner’s tools
D 4.3: Benchmarking report as tested on the available infrastructure
D 4.4: API and runtime for IO using persistent storage
D 4.5: Framework development and release
WP 5 Algorithmic extensions of MLMC (UQ)
WP 6 Optimization under uncertainties
D 5.2: Release of ExaQUte MLMC python engine
D 5.3: Report on theoretical work to allow the use of MLMC with adaptive mesh refinement
D 5.4: Report on MLMC for time dependent problems
D 5.5: Report on the application of MLMC to wind engineering applications
D 6.1: Deterministic Optimization Software
D 6.2: Report on the calculation of stochastic sensitivities
D 6.3: Report on stochastic optimization for simplified problems
D 6.4: Report on stochastic optimization for unsteady problems
D 6.5: Report on stochastic optimization for wind engineering
WP 7 Application to robust shape optimization of structures under wind loads
WP 8 Dissemination and exploitation
D 7.1: Delivery of geometry and computational model
D 7.2: Finalization of “deterministic” verification and validation tests
D 7.3: Report on UQ results and overall user experience
D 7.4: Final report on Stochastic Optimization results
D 8.1: Data Management Plan
D 8.2: Report on dissemination activities
D 8.3 : Exploitation plan
D 8.4 : Report on dissemination activities
WP 9 Project management
D 9.1: Periodic technical, administrative and financial report
D 9.2: Periodic technical, administrative and financial report