Abstract. Four perturbation theories presently used to compute asteroid proper elements are reviewed, and their results are briefly discussed (Milani and Knezevic, 1990, for low to moderate eccentricity/inclination main belt objects; Lemaitre and Morbidelli, 1994, for high e, J objects; Milani, 1993, for Trojans;Schubart, 1982, 1991. The most important recent improvements are described, in particular those pertaining to the upgrades of the previous analytic and semianalytic solutions. The dynamical structure of the asteroid main belt, as defined by the low order mean motion resonances and by linear and nonlinear secular resonances, is considered from the point of view of the effects of these resonances on the accuracy and/or reliability of the computation of proper elements and on the reliability of the identification of asteroid families.
What are the proper elements and what they are used for?Asteroid proper elements are quasi-integrals of motion, stable over very long intervals of time; they represent a sort of "average" characteristics of motion in the sense that they result from a procedure of elimination of short-and long-periodic perturbations. The proper elements are used for two main purposes : (i) they serve as parameters for classification of asteroids into families; (ii) they are employed for studies of the dynamical structure of the asteroid belt.Asteroid families are groupings of asteroids in the phase space of orbital elements. They were formed by catastrophic breakup of parent bodies, at unknown times in the remote past, and under circumstances (such as geometry of collision, physical properties of the parents, etc.) that are not yet fully understood. The parent body at the breakup might have had osculating elements significantly different from the corresponding proper values; since we do not know the epoch of the collision, the proper elements (due to their constancy in time) become the only tool to reliably recognize families, in particular the small ones consisting of a few members. Until recently, the very identification of the families and their memberships were not very reliable : the proper elements used for this identification were not accurate enough, and/or the classification methods suffered from subjectivity, biases, uncertainties due to insufficient data, etc. The situation is now greatly improved in both respects; the review of the current state of the art regarding proper elements and family classification is the subject of this and some other papers in this book (see e.g. Zappalà and Cellino, Milani, Burbine and Binzel).The differences of proper orbital elements between family members are related to their relative velocities, averaged over the unknown quantity, the age of the family. In order to reliably recognize the families, the changes in the proper elementsover time spans of the order of the family ages-must not exceed an amount given by the typical relative velocities of the family members. The relative velocities of family members are known to be of the order of 100 m/s; f...