On dimension and existence of local bases for multivariate spline spaces (Q1198155)

The authors investigate polynomial splines in \(k\) variables defined on a \(k\)-dimensional triangulation. Generalizing the bivariate approach, suitable Bernstein-Bézier techniques are introduced to deal with the problems of dimension and of construction of local bases for spaces of \(k\)-variate splines of smoothness \(r\) and total polynomial degree \(d\). For the case of continuous \(k\)-variate piecewise polynomials of degree \(d\), i.e. \(r=0\), the dimension of the spline space is given and a local basis is described. For \(r>0\), the analysis of the smoothness becomes much more difficult. Detailed results are established for trivariate splines on special partitions which in turn are used to prove a dimension formula and find a minimally supported basis for trivariate spline spaces on arbitrary tetrahedral partitions if \(d>8r\).