Locally tensor product functions in algebraic grid optimization

This paper deals with an a posteriori local improvement of numerical grids generated trough a new algebraic mixed scheme. The improvement is achieved by an optimization strategy where only a small number of new variables is used to provide an ‘‘optimal’’ algebraic grid. The proposed new mixed scheme is based on a special family of functions that is tensor product functions in sub-regions, and it controls the position of the grid points by moving a subset of control points of the scheme.