Asteroid identification at discovery

We present a novel method for the search of linkages among astrometric observations of asteroids, that is, tentative identifications among asteroids observed. Having two different master sets of asteroid observations each containing a number of separate subsets, we define a linkage as a pair of subsets residing in separate master sets that can be tied together with an orbit for given observational errors. To find linkages among a wealth of observations we use an efficient stepwise filtering approach. First, we start with what we call phase-space address comparison. The first step substantially reduces the initially huge amount of pairs by requiring that pairs to be subjected to further analysis have similar geocentric spherical coordinates at common epochs (for example, at three epochs). Second, we search for orbits for each of the selected pairs of subsets. Succeeding in the effort proves that a linkage exists. If there are contradictions among linkages found-for example, a single subset being linked to several mutually exclusive subsets-additional new or archive observations are usually needed to discard erroneous linkages. The new method is built on six-dimensional statistical orbital inversion (Ranging), and is therefore particularly suitable for analyzing objects with the shortest observational arcs, that is, newly discovered asteroids (and comets). Results from extensive and successful tests on simulated survey observations are presented and discussed. Theoretical and empirical scaling results show that the method is applicable to future large-scale surveys that will increase the rate of asteroid discovery by at least two orders of magnitude. The successful linking of faint single-night observation sets obtained with the Very Large Telescope are briefly reviewed.