The Laboratory for Scientific Computing covers a wide range of applications. However, the approach to simulating each physical situation usually requires very similar numerical and computational techniques. The majority of research projects under investigation in the Laboratory can be seen as the solution of a continuum problem on a discrete computational space.
Many of the research areas can be classed as fluid dynamical problems, which can be modelled as hyperbolic systems (for conservative fluid flow) coupled with chemical or geometrical effects, which enter through source terms.
All our computational simulations have a solid grounding in efficient, stable, and robust numerical discretizations of the physical models we wish to solve. Not all of the following numerical methods are required by each research area, but a choice will be made of which are the most suitable techniques for a particular problem.
- High Resolution Solvers
- Adaptive Mesh Refinement
- Cut-Cell Methods
- Eulerian Multi-Material Simulation
High-Performance Computing Techniques
Once the most appropriate numerical techniques have been chosen for a particular problem, they must be applied efficiently. This requires an understanding of modern computer architecture, and the techniques that can be used to extract maximum performance from it:
- Computing with GPUs (Graphics Processors)
- GPUs applied to Gas Dynamics
Following are some of the research areas currently under investigation by members of the Laboratory.
- Deflagration to Detonation Transition
- Detonations in Heterogeneous Explosives
- Flow in Porous Media
- Numerical Relativity
- Void Collapse in Multiphase Flow
- Semiconductor device simulations
- Earth-system modelling: AMR for global atmospheric models.