A note on packing random intervals with varying density
It is known that given N random subintervals of [0, 1], one can find a disjoint subcollection that covers all of [0, 1] except a set of length about (log N)2/N. We investigate what happens when the distribution of the intervals is biased to favor shorter intervals or intervals close to the endpoints of [0, 1]. Quite surprizingly, the order (log N)2/N is very robust.