Compressed Bit Vectors Based on Variable-to-Fixed Encodings

Abstract

We consider practical implementations of compressed bitvectors, which support rank and select operations on a given bit-string, while storing the bit-string in compressed form. Our approach relies on variable-to-fixed encodings of the bit-string, an approach that has not yet been considered systematically for practical encodings of bitvectors. We show that this approach leads to fast practical implementations with low redundancy (i.e., the space used by the bitvector in addition to the compressed representation of the bit-string), and is a flexible and promising solution to the problem of supporting rank and select on moderately compressible bit-strings, such as those encountered in real-world applications.