On a norm equivalence in \(L^2(\mathbb S^2)\) (Q849430)

For the unit sphere \(\mathbb{S}^2\) in three space dimensions \(\mathbb{R}^3\) an inner product is defined in [\textit{D. Roşca}, Math. Comput. 74, No. 252, 1803--1829 (2005; Zbl 1091.42025)] using a convex polyhedron with triangular faces. All its vertices are supposed to be situated on the sphere and no face contains the origin, the latter being inside the polyhedron. The inner product on the polyhedron induces an \(L^2\)-norm which is shown to be equivalent to the usual \(L^2\)-norm on the sphere. Therewith wavelets, originally constructed on a bounded domain in \(\mathbb{R}^2\), can be lifted onto the sphere, while maintaining their numerical behaviour on the plane. In the current work a more general construction is presented where the polyhedron is no longer inscribed in the sphere. Likewise a norm equivalence holds and optimal bounds are derived. Furthermore it is shown that scaling or rotation of the polyhedron around the origin do not alter these optimal bounds. Finally, possible applications to construct orthogonal wavelets on the sphere are presented which rely on the orthogonality with respect to the inner product associated to the polyhedron. Hence the wavelet decomposition and reconstruction algorithms on the sphere inherit the orthogonality and local support properties of the wavelets in the plane, while avoiding distortions and singularities around the poles.