Next: One-Dimensional Orthogonal Polynomials
Up: Spectral/hp Methods on Polymorphic
Previous: Algorithm For Connecting Prisms
This chapter is devoted to outlining the polynomial bases that we use
to represent a function defined on an element. A desirable basis has
three properties: (i) it should be orthogonal or nearly orthogonal
under a convenient norm, (ii) well behaved at high polynomial order,
and (iii) it should be computationally efficient to take inner
products with this basis if it is to be used in a Galerkin
framework. The bases we present are designed to satisfy these
conditions. The computational complexity of taking inner products has
been minimized for these hierarchical bases by insisting that each
basis function is a tensor product of one-dimensional shape functions.
T. Warburton
10/24/1998