Fast evaluation of system matrices w.r.t. multi-tree collections of tensor product refinable basis functions

An algorithm is presented that for a local bilinear form evaluates in linear complexity the application of the stiffness matrix w.r.t. a collection of tensor product multiscale basis functions, assuming that this collection has a multi-tree structure. It generalizes an algorithm for sparse-grid index sets [R. Balder, Ch. Zenger, The solution of multidimensional real Helmholtz equations on sparse grids, SIAM J. Sci. Comput. 17 (3) (1996) 631-646] and it finds its application in adaptive tensor product approximation methods.