Conquering the pre-computation in two-dimensional harmonic polynomial transforms
From MaRDI portal
Eigenvalues, singular values, and eigenvectors (15A18) Orthogonal polynomials and functions in several variables expressible in terms of special functions in one variable (33C50) Orthogonal polynomials and functions associated with root systems (33C52) Spherical harmonics (33C55) Linear symmetric and selfadjoint operators (unbounded) (47B25) Jacobi (tridiagonal) operators (matrices) and generalizations (47B36)
Abstract: We describe a skeletonization of the spherical harmonic connection problem that reduces the storage and pre-computation to superoptimal complexities at the cost of increasing the execution time by the modest multiplicative factor of . One advantage of accelerating the spherical harmonic connection problem over accelerating synthesis and analysis is that neighbouring layers (in steps of two) may be expanded in eachother's bases. The proposed skeletonization maximizes this interconnectivity by overlaying a dyadic partitioning on the connection problem. We derive the symmetric-definite banded generalized eigenvalue problem required to accelerate spherical harmonic transforms. We also include a full analysis of the weighted normalized Jacobi connection problem with applications to fast harmonic polynomial transforms on the disk, triangle, rectangle, deltoid, wedge, and any other geometry with a bivariate analogue of Jacobi polynomials.
This page was built for publication: Conquering the pre-computation in two-dimensional harmonic polynomial transforms
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6294224)
