Mixed genetic strategies for point pattern reconstruction with substantially incomplete information

Two dimensional point patterns with n points can be uniquely represented by as few as an appropriately chosen 2n-3 out of a total of n(n-1)/2 interpoint distances. This paper first considers reconstruction of such patterns when the distances measurements are exact, and then considers the more general, and more interesting, problem of reconstruction when the distances contain possible measurement errors. Reconstructions using genetic algorithms (GA) are compared to those using multidimensional scaling (MDS) and it is shown that when the number of measured distances approaches the theoretical minimum of 2n-3, mixed strategies based on GA´s prove most efficient