Computer programs for fitting ultrametric and additive trees to proximity data by least squares methods
Geert De Soete · Behavior Research Methods, Instruments, & Computers · 1984
LSULT (Least Squares ULtrametric Tree analysis) and LSADT (Least Squares ADditive Tree analysis) are computer programs that implement the least squares algorithms proposed by De Soete (1983, 1984b) for fitting ultrametric and additive trees to proximity data. An ultra~etric tree is a rooted tree in which a nonnegative weight IS attached to each node such that: (1) the terminal nodes have zero weight, (2) the largest weight is assigned to the root, and (3) the weights attached to the nodes on the path from any terminal node to the root constitute a strictly increasing sequence. The distance between any two disti.nctnodes is defined as the maximum of the weights assignedto the nodes on the unique path connecting the two nodes. It is well-known that these distances satisfy the ultrametric inequality: