arXiv · 2611.14764
Orthogonal Splines Associated to Root Systems and Applications to Function Approximation From Generalized Derivative Samples
Abstract
Let $\mathcal{S}(X)$ be the principal shift invariant space generated by a box spline $B_X$ with $d$-dimensional direction set $X$ and associated differential operator $D_X$. Assuming that each direction in $X$ appears with even multiplicity i.e $X = [X_0|X_0]$, we may endow $\mathcal{S}(X)$ with a special inner product $\langle \cdot , \cdot \rangle_X$ which is similar to a homogeneous Sobolev-like inner product of order $|X_0|$. Then, for the special case when the directions in $X$ are selected from a root system, we introduce bi-orthogonal systems for $\mathcal{S}(X)$. Within any fixed bi-orthogonal system, each basis function is compactly supported albeit the diameter of the supports of the basis functions is not uniformly bounded. On the other hand, the image of any bi-orthogonal system under $D_X$ (which is necessarily finite and discrete) is uniformly finite. In other words, the basis elements, while having non-sparse box-spline representation, are uniformly sparse in their truncated-power representation. Using the Sobolev-like inner product and an aforementioned biorthogonal system for $\mathcal{S}(X)$, the analysis-synthesis operator from smooth compactly supported continuous functions into $\mathcal{S}(X)$ is well-defined. Using integration by parts and the orthogonality satisfied by the bases under $\langle \cdot , \cdot \rangle_X$, this operator is a local discrete quasi-interpolation operator. When $X$ comes from an irreducible root system, we can slightly modify these quasi-interpolants to ensure they have optimal approximation order. In other words, if $f$ is sufficiently smooth and compactly supported, we can optimally approximate $f$ from a collection of generalized derivative samples induced by one of our bases.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Dionel Jaime. 2026-11-10. Orthogonal Splines Associated to Root Systems and Applications to Function Approximation From Generalized Derivative Samples. https://searcharxiv.com/abs/2611.14764
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.