Codes created by the CRUNCH group:
The group has been developing several codes mainly for its own use.
Most of the codes are designed to be used on massively parallel distributed
memory machines and use either standard message passing libraries
PVM) or in some
cases the native message passing calls for better performance (NX on the
Paragon, SMA on the T3D). New parallel code development is done using MPI
as it is clearly becoming the accepted standard.
Anyone interested in using these codes in one's own research should
Prof. George Karniadakis for further information.
PRISM and derivatives
- (Non-conforming) Spectral Element Codes:
- 2D and 3D versions
- Assuming a homogeneous z-direction, uses Fourier transforms for 3D
- Used for:
- Turbulence Calculations
- Flow-Structure Interactions
- MagnetoHydroDynamics Calculations
- Ron D. Henderson
- David J. Newman
- Catherine H. Crawford
Prism 3D has been ported to:
- (Conforming) Cylindrical Coordinate Spectral Element Code for axi-symmetric geometries:
- "2D" version for axi-symmetric flows
- 3D version uses Fourier transforms for the angular direction.
- David J. Newman
- (Capabilities Summary)
- Triangular/Tetrahedral Spectral Elements
- Unstructured Meshes
- New "polymorphic element" C++ code (developed by Tim Warburton) uses triangles/rectangles for 2D calculations. 3D version uses a mix of tetrahedra, prisms and hexahedra.
- PRISM-like approach for another 3D code using Fourier expansion in the z-direction.
- Fully 3D parallel code for both Nektar and "polymorphic" Nektar.
- Spencer J. Sherwin
- Timothy C.E. Warburton
Fourier Nektar 3D (and new polymorphic element code) are MPI based and is being used on:
- Heat Transfer in complex micro-geometries.
- Ali Beskok
- Incompressible Vorticity-Velocity Algorithm
- 2D and 3D (MPI version under development)
- James Trujillo
Stochastic Nektar code to model uncertainty in fluid flows,
- Dongbin Xiu
- Wiener-Askey polynomial chaos expansion in random space:
- Hermite-Chaos for Gaussian random inputs;
- Laguerre-Chaos for Gamma random inputs;
- Jacobi-Chaos for Beta random inputs;
- Legendre-Chaos for Uniform random inputs (special case of Jacobi-Chaos).
- Chaos expansions can be multidimensional according to the random inputs.
- Utilize high-order spectral/hp element method on unstructured hybrid
- Parallel version for s-Nektar 2D.
- Various postprocessing routines, eg, s-Vort2d etc for vorticity mean,