Fingerprint Filters Are Optimal
By: William Kuszmaul, Jingxun Liang, Renfei Zhou
Potential Business Impact:
Makes computer searches faster and smaller.
Dynamic filters are data structures supporting approximate membership queries to a dynamic set $S$ of $n$ keys, allowing a small false-positive error rate $\varepsilon$, under insertions and deletions to the set $S$. Essentially all known constructions for dynamic filters use a technique known as fingerprinting. This technique, which was first introduced by Carter et al. in 1978, inherently requires $$\log \binom{n \varepsilon^{-1}}{n} = n \log \varepsilon^{-1} + n \log e - o(n)$$ bits of space when $\varepsilon = o(1)$. Whether or not this bound is optimal for all dynamic filters (rather than just for fingerprint filters) has remained for decades as one of the central open questions in the area. We resolve this question by proving a sharp lower bound of $n \log \varepsilon^{-1} + n \log e - o(n)$ bits for $\varepsilon = o(1)$, regardless of operation time.
Similar Papers
Smaller and More Flexible Cuckoo Filters
Data Structures and Algorithms
Makes computer memory smaller and faster.
Succinct Dynamic Rank/Select: Bypassing the Tree-Structure Bottleneck
Data Structures and Algorithms
Makes computer lists change super fast, using less space.
Improved Online Sorting
Data Structures and Algorithms
Sorts numbers faster, saving space and time.