Approximate multiplication in adaptive wavelet methods (Q352734)

The authors present a new approximate matrix-vector multiplication which better utilizes actual decay of the entries in the wavelet stiffness matrix and in the vector. It is shown that this multiplication algorithm is asymptotically optimal in the sense that storage and number of floating point operations remain proportional to the problem size. For some wavelet stiffness matrices, the new multiplication technique is compared with corresponding techniques of \textit{A. Cohen}, \textit{W. Dahmen}, and \textit{R. DeVore} [Math. Comput. 70, No. 233, 27--75 (2001; Zbl 0980.65130)] and of \textit{T. J. Dijkema}, \textit{C. Schwab}, and \textit{R. Stevenson} [Constr. Approx. 30, No. 3, 423--455 (2009; Zbl 1205.65313)].