{"id":"arxiv/2602.06284","url":"https://arxiv.org/abs/2602.06284","tags":["math.NA"],"score":0,"title":"Geometric Kernel Interpolation and Regression","domain":"math.NA","source":"arxiv","authors":["Patrick Guidotti"],"reviews":0,"subject":"math.NA","summary":"Exploiting the variational interpretation of kernel interpolation we exhibit a direct connection between interpolation and regression, where interpolation appears as a limiting case of regression. By applying this framework to point clouds or samples of smooth manifolds (hypersurfaces, in particular), we show how fundamental geometric quantities such as tangent plane and principal curvatures can be computed numerically using a kernel based (approximate) level set function (often a defining function) for smooth hypersurfaces. In the case of point clouds, the approach generates an interpolated…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-02-06","reactions":{}}