The obvious way to speed up computations is to use a faster computer. An alternative is to improve the efficiency of the computations. CWI researcher Marc van Raalte worked on techniques that accelerate the numerical approximation of partial differential equations. They can for example be used in simulations of gas and fluid flows or the diffusion of heat. On December 6, van Raalte defended his PhD thesis at the Universiteit van Amsterdam.
In his thesis van Raalte combines discontinuous Galerkin methods with multigrid algorithms. Individually, these techniques have proved themselves capable of speeding up many types of approximations. But until now, no elegant way to combine them existed. Van Raalte also developed a computationally efficient way to deal with curved boundaries like ship's hulls in computations. Research groups from the U.S. including NASA have already shown interest in Van Raalte's work.
More information can be found on MAS2's website.
