In this talk, we will present several techniques for the construction of multiple levels of unstructured meshes that approximate a scattered data set at different levels of detail. We will focus on two techniques. The first approximates the data set by a linear spline function over a tetrahedral grid and concentrates on the reduction (collapse) of individual tetrahedra. The second method uses a "meshless" representation of the data set, creating clusters of data points and creating the different levels by approximating the clusters by a single data point.

Analysis of the error in the approximations generated by each method will be presented. These methods can be used to generate hierarchical descriptions of the data set suited for the efficient visualization of the data set at various levels of detail.

Snacks will be provided.

See Conundrum Talks for more information about this series.