Sharp Amplitude Limits for Robust Multivariate Synthesis: Gaussian and Nonquadratic Kernels
We determine the sharp cost of achieving a prescribed sign pattern with translated kernels under a unit total variation budget. The observations may be finitely many labeled points or two disjoint compact sets. For Gaussian kernels and smooth homogeneous potentials, the leading high-margin cost is determined by a convex margin problem and a stationary barycentric transport problem. For polyhedral potentials, face profiles give the sharp coefficient and quantitative error bounds. We also give exact results for small Gaussian configurations and quantify the cost of restricting the number of sources on a sphere.
Authors
- Akihiro Koide (ORCID: https://orcid.org/0009-0002-8391-3480)
Publication Details
- Journal
- Figshare
- Published
- 2026-11-13
- DOI
- https://doi.org/10.6084/m9.figshare.34999708
- Primary Topic
- Geometric Analysis and Curvature Flows
- Type
- preprint