11/22/2017 created by Vincent Heuveline
HiFlow3 is now ready for Uncertainty Quantification (UQ)
Version 2.0 of HiFlow3 provides major and essential new features in several of its modules: The mesh module provides extended support for adaptively refined meshes by means of the p4est library. Furthermore, pyramid elements are now supported for 3D geometries. The DoF/FEM modules now provide the ability for Discontinuous Galerkin (DG) discretizations. The linear algebra module ships with new solvers and preconditioners (Schur complement and energy-efficient geometric multigrid) as well as interfaces to further third-party libraries (Hypre, PETSc, SLEPc). Furthermore, new (parallel) matrix and vector implementations for Polynomal Chaos Expansion (PCE) based discretizations are available. Problems related to Uncertainty Quantification (UQ) for PDEs can be solved using these techniques. Finally, tools for goal oriented adaptivity are now available in the context of high performance computing.
The new features are documented in several new examples and tutorials.
- New mesh module mesh_pXest: capable of local mesh refinement, based on p4est library
- New mesh element: pyramid
- Support for discontinuous Galerkin
- Schur complement solver and preconditioner
- Energy-efficient parallel geometric multigrid solver and preconditioner
- Interfaces to Hypre, PETSc and SLEPc libraries
- New linear algebra structure: matrix and vector structure for Polynomial Chaos Expansion
- Tools for goal oriented adaptivity: patch interpolation, fe space interpolation, dynamic mesh handler
- Aortic Blood Flow Simulation
- Error Estimation on Convex Bent Domains for the Poisson Equation