SELF-ADJUSTING AUGMENTED SEARCH-TREES

We consider the problem of maintaining a dynamic weighted binary search tree augmented with a secondary search structure. Although we show that partial rebuilding cannot simultaneously achieve optimal search and update times, we introduce a new technique related to partial building called weighted partial rebuilding, which supports optimal worst-case search times (within a constant factor) for primary keys, O(log n) amortized update times, and efficient amortized reweight times. We also give an example application.