Blending second-order maximum-entropy approximation schemes and B-Splines (Poster)

ADRIAN ROSOLEN; DANIEL MILLÁN; MARINO ARROYO

Cardona, Barcelona

Workshop; 9th Workshop on Numerical in Applied Science and Engineering 2010; 2010

LaCàN - Universitat Politècnica de Catalunya

The application of computational geometry technologies like B-Splines or NURBS in the field of numerical simulations has been proposed in recent years, mainly incited by the aim of bringing together CAD and FEA. A considerable advantage of these approximation schemes is the possibility to describe exactly complex geometries, while the main drawback is the need of structured meshes. In spite of being endowed with the properties of convex geometry (smoothness, non-negativity, variation diminishing property), maximum-entropy approximants do not allow an exact geometric description, but present flexibility to work with unstructured sets of points. The similar nature of these approximants, and the opposite character of their drawbacks and advantages, motivate to mix them. Here, two methods for blending second-order maximum-entropy (SME) approximants and B-Splines are proposed, and preliminary results are shown. The Centroidal Voronoi tesselation technique is also illustrated.