Here we show how the distribution of semimajor axes, eccentricity, and inclination of the family members disperses with time.   For the simulations with real members of Adeona and Gefion, the red line represent the initial distribution (in the first row we report the identified members as from Cellino's file, while in the second row we report the members identified from David Nesvorny's program with a cutoff of 80 m/s (in this case the number of observed asteroids is normalized to the number of particles in the simulations)), the green line the results at t=300 My, and the blue line the results at t=500 My.
For the simulations with synthetic families, the red, green, and blue lines have the same meaning of above and the black line represents the initial distribution of synthetic family members.
 
 

For the simulation with real members of Adeona, we observe that the major effect was to introduce spreading in e and i via interaction with resonances and microresonances, with small changes in a,  whose distribution (with some exceptions) is conserved (see paragraph about "triangle diagrams").   About the synthetic family's integrations, the simulation with close encounters show that the spreading in a is increased, but can't account for to the observed distribution according to Cellino's file, and much less for the distribution computed from the results of D. Nesvorny's program with a cutoff at v=80 m/s.  We also observe that the center in e and i of the family is not the same for the two distributions.   Since in generating the synthetic family we used the members of Adeona as from Cellino's file, that explain the offset in e and i  of the simulated initial conditions with Nesvorny's distribution at v=80 m/s.   Yarkovsky's results at t~300 My are able to match the current distribution in a, (see section on families' ages) but not in e and i (which is probably due to the fact that the Yarkovsky effect is mainly acting on a, so the final distribution  in e and i is probably due to the too conservative  value of velocity at infinity we choose to generate our synthetic family).   Note that since the drift rates in a are higher for the Yarkovsky effect than for close encounters with large asteroids, this mechanism is able to "feed" resonances at a faster pace, so that the final distribution at t~500 My in e and i is have a higher values of sigma than the corresponding results from close encounters'simulation.
 
 

Many of the observation made for Adeona also apply to the case of Gefion, so we are not going to repeat them.  We just observe that the offset between Cellino's and Nesvorny's center of the family is now in a and partly in e, and that Yarkovsky's simulation do not match the current distribution as from Nesvorny's program even at t =500 My (see section on families' ages) .  The simulation with close encounters had the largest variation of sigma_a, with almost 30% of the original value.   Close encounters alone were able to replicate 46% of sigma_a as from Cellino, and 28% of sigma_a as from Nesvorny's code (the distribution at t =0 was 35% and 21%, respectively).
 
 

NOT YET AVAILABLE.