[Paper Review] On Dynamic Range Reporting in One Dimension

Christian Worm Mortensen, Rasmus Pagh+1 more|ArXiv.org|Feb 5, 2005

Algorithms and Data Compression11 references39 citations

TL;DR

This paper presents a dynamic data structure for one-dimensional range reporting that achieves O(lg w) update time and O(lg lg w) query time on a word RAM with w-bit words, significantly improving upon predecessor search bounds. The solution uses a novel recursion scheme—more powerful than van Emde Boas recursion—applied across every path in a binary trie, enabling exponentially faster queries while maintaining optimal O(n) space via a new dynamic perfect hashing scheme with sublinear space usage.

ABSTRACT

We consider the problem of maintaining a dynamic set of integers and answering queries of the form: report a point (equivalently, all points) in a given interval. Range searching is a natural and fundamental variant of integer search, and can be solved using predecessor search. However, for a RAM with w-bit words, we show how to perform updates in O(lg w) time and answer queries in O(lglg w) time. The update time is identical to the van Emde Boas structure, but the query time is exponentially faster. Existing lower bounds show that achieving our query time for predecessor search requires doubly-exponentially slower updates. We present some arguments supporting the conjecture that our solution is optimal. Our solution is based on a new and interesting recursion idea which is "more extreme" that the van Emde Boas recursion. Whereas van Emde Boas uses a simple recursion (repeated halving) on each path in a trie, we use a nontrivial, van Emde Boas-like recursion on every such path. Despite this, our algorithm is quite clean when seen from the right angle. To achieve linear space for our data structure, we solve a problem which is of independent interest. We develop the first scheme for dynamic perfect hashing requiring sublinear space. This gives a dynamic Bloomier filter (an approximate storage scheme for sparse vectors) which uses low space. We strengthen previous lower bounds to show that these results are optimal.

Motivation & Objective

• To design a dynamic data structure for range reporting in one dimension that achieves faster query times than predecessor search structures.
• To overcome the inherent trade-off between update and query time in predecessor search, where faster queries require doubly-exponential updates.
• To achieve optimal O(n) space usage in dynamic range reporting, which had not been previously achieved with such fast query times.
• To develop a new dynamic perfect hashing scheme that uses sublinear space, solving a key problem for sparse vector representation.
• To prove optimality of the proposed solution through new lower bounds for the greater-than function and related problems.

Proposed method

• The core technique uses a nontrivial, van Emde Boas-like recursion on every root-to-leaf path in a binary trie, rather than simple halving, to enable faster query resolution.
• The method applies this recursion to the bit-probe complexity of the greater-than function, which is then generalized to dynamic range reporting by applying it across all paths in the trie.
• A new dynamic perfect hashing scheme is introduced that uses sublinear space, enabling efficient storage of sparse vectors and supporting the Bloomier filter problem.
• The data structure maintains a hierarchical trie with multiple levels of recursion, where each level uses a Bloomier filter to track branching ancestors and enable fast navigation.
• Querying involves a binary search over a sequence of recursive structures, with checks at each level to verify correctness of ancestor pointers, ensuring correctness even when Bloomier filters return arbitrary results for invalid nodes.
• Lower bounds are derived using the sunflower lemma and Fredman’s proof technique, showing that any solution with suboptimal query time must have exponentially slower updates.

Experimental results

Research questions

• RQ1 Can range reporting in one dimension be solved with query time significantly faster than predecessor search, while maintaining efficient update times?
• RQ2 Is it possible to achieve O(lg lg w) query time with O(lg w) update time in the word RAM model, despite known lower bounds for predecessor search?
• RQ3 Can a dynamic perfect hashing scheme be constructed that uses sublinear space while supporting correct operations on non-null keys?
• RQ4 What is the optimal trade-off between update and query time for dynamic range reporting in one dimension?
• RQ5 Are the proposed bounds for the greater-than function and range reporting problem tight, and can they be proven optimal via information-theoretic lower bounds?

Key findings

• The paper achieves O(lg w) update time and O(lg lg w) query time for dynamic range reporting, which is exponentially faster than predecessor search for the same update time.
• The proposed solution uses a novel recursion scheme that is more powerful than the van Emde Boas recursion, enabling faster query resolution across all paths in the trie.
• A new dynamic perfect hashing scheme is developed that uses sublinear space, solving the problem of space-efficient sparse vector storage for the Bloomier filter problem.
• The data structure uses O(n) words of space, achieving optimal space usage for dynamic range reporting.
• Lower bounds are proven showing that any solution with O(lg lg w) query time must have update time at least Ω(2^{w^{1-ε}}) for predecessor search, confirming the optimality of the trade-off in this work.
• The results are shown to be optimal via a combination of the sunflower lemma and Fredman’s proof technique, establishing tight bounds for the greater-than function and related problems.