Depending on the relative sizes of the sets and the number of times the intersections are computed, it might be more efficient to store the sets as ordered list

Depending on the relative sizes of the sets and the number of times the intersections are computed, it might be more efficient to store the sets as ordered lists and do a linear merge pass to compute the intersection.