LZ

LZ is the abbreviation for Lempel-Ziv, a family of lossless data compression algorithms developed by Abraham Lempel and Jacob Ziv in the 1970s. The family includes several variants, most notably LZ77, LZ78, and LZW, each using a different approach to model redundancy in data.

LZ77, introduced in 1977, uses a sliding window of previously seen data as a dictionary and encodes matches as a pair (distance, length) plus a following symbol. It is particularly effective on repetitive data and forms the basis for many modern compression schemes such as DEFLATE (used in ZIP and GZIP).

LZ78, introduced in 1978, builds a dictionary of previously seen sequences and encodes input as a sequence of dictionary indices plus new symbols as they appear, enabling constant growth of the dictionary.

LZW, proposed by Terry Welch in 1984, is a practical variant of LZ78 that dynamically builds the dictionary and outputs codes for dictionary entries rather than raw data. It achieved widespread use in the GIF image format and in some TIFF and other systems; patent activity around LZW influenced its adoption until expiration in the 2000s.

LZ-based schemes also include LZSS, a modification of LZ77 that improves compression efficiency by using fixed-length codes and explicit escape codes.

In modern practice, many standards blend LZ principles with entropy coding; for example, DEFLATE combines LZ77-style matching with Huffman coding. LZ remains foundational in data compression theory and implementations.