Description
Summary
I propose to add Feel++ package.
Feel++ is an Open-Source C++ library designed to solve a wide range of partial differential equations (PDEs) using advanced Galerkin methods. These methods include the finite element method (FEM), spectral element method, discontinuous Galerkin methods, and reduced basis methods.
Rationale
Feel++ is part of the software stack of the French exascale initiative NumPEx
Description
Feel++ is an Open-Source C++ library designed to solve a wide range of partial differential equations (PDEs) using advanced Galerkin methods. These methods include the finite element method (FEM), spectral element method, discontinuous Galerkin methods, and reduced basis methods.
Feel++ is optimized for high-performance computing, enabling seamless parallel computing on large-scale systems, ranging from desktop machines to supercomputers with tens of thousands of cores. The library supports multi-physics simulations and provides a modular structure to simplify the development of applications.
Key Features:
- Toolboxes: Predefined toolboxes for common PDE problems, enabling faster development and deployment of complex simulations. These toolboxes include:
- Coefficient Form PDEs: For general PDEs in coefficient form.
- Computational Solid Mechanics: For simulations of solid mechanics problems.
- Computational Fluid Mechanics: For solving fluid dynamics problems.
- Heat Transfer: For heat conduction and convection problems.
- Heat & Fluid: Coupled heat and fluid dynamics simulations.
- Fluid Structure Interaction (FSI): For problems involving interactions between fluid flow and structural mechanics.
- Electric: For solving electric field and current problems.
- Thermo-Electric: For coupled thermoelectric simulations.
- Maxwell: For simulating electromagnetic field problems based on Maxwell's equations.
- Hybridized Discontinuous Galerkin: For high-order accurate discontinuous Galerkin methods with hybridization techniques.
- Model Order Reduction (MOR): Provides efficient techniques for reducing the computational complexity of large-scale simulations, enabling faster simulations while maintaining accuracy.
- Python Wrappers: Python bindings allow for seamless integration of Feel++ functionalities into Python-based workflows, enabling rapid prototyping, interactive simulations, and data visualization.
- High-Performance Computing (HPC): Fully optimized for distributed (MPI) and shared-memory (multithread) parallelism, Feel++ can efficiently scale from small multi-core systems to large HPC clusters.
- C++ Standards: Support for C++17, C++20, and C++23 standards, allowing users to leverage modern language features and optimizations.
- **Extensive Scientific Libraries Integration**: Feel++ integrates with major scientific computing libraries, including PETSc, SLEPc, Boost, FFTW, and others, to extend its functionality and scalability.
Optional Variants:
- +toolboxes: Enable specialized toolboxes for common PDEs, such as fluid mechanics, solid mechanics, and coupled heat and fluid simulations.
- +mor: Enable Model Order Reduction (MOR) for computationally expensive large-scale systems.
- +python: Enable Python bindings to allow Python-based interaction with Feel++.
- +quickstart: Enable quickstart examples for easier onboarding and usage.
- +tests: Enable tests to verify the integrity of the Feel++ installation.
- cpp17/cpp20/cpp23: Choose the C++ standard (C++17, C++20, or C++23) for compilation.
Additional information
No response
General information
- I have searched the issues of this repo and believe this is not a duplicate