Geometrically smooth spline bases for data fitting and simulation (Q2174076)

The authors present a new method to construct basis functions for \(G^1\)-continuous spline spaces supported on quadrangular meshes. Their approach builds on \(G^1\)-spline constructions for shapes with complex topology studied in [\textit{B. Mourrain} et al., Comput. Aided Geom. Des. 45, 108--133 (2016; Zbl 1418.41010); \textit{A. Blidia} et al., ibid. 52--53, 106--125 (2017; Zbl 1366.65012)]. The goal is to obtain a better approximation of a surface starting from a coarse representation of the shape that captures its topology and principal geometric features. If the coarse model is a mesh, a classical strategy is to refine the mesh by splitting some of its faces. This approach yields piecewise linear representations of the shape, which may require several levels of subdivisions in regions with high curvatures. Alternatively, instead of splitting the coarse piecewise linear model, the authors propose to construct a \(G^1\)-spline space and increase the degree of the representation on each quadrangular face. The so-called geometric continuity, or \(G^1\)-splines, is a technique that consists of gluing parametric surface patches along their common boundary, with continuity constraints on the tangent planes. The authors adapt the construction of minimal determining sets [\textit{P. Alfeld}, J. Comput. Appl. Math. 119, No. 1--2, 13--27 (2000; Zbl 0966.65017)] and present a method to efficiently deduce basis functions using local linear operations associated to the edges of the mesh. The results are applied to fitting problems of point clouds, and to the solution of diffusion equations, following an isogeometric analysis approach.