LZMA Algorithm

LZMA is a lossless data compression algorithm that shrinks a file's size while preserving every bit needed to reconstruct the original exactly on decompression. It combines a dictionary-based compression scheme in the lineage of LZ77, but using a much larger dictionary than LZ77 implementations typically allow, with range encoding that assigns each bit a probability drawn from a context specific to its position within the data's internal representation, rather than treating each byte generically, which lets it predict and encode likely bits more efficiently. Igor Pavlov began developing LZMA in 1998 and built it into his 7-Zip archiver starting in 2001; the reference implementation was originally released under a mix of the GNU Lesser General Public License and the Common Public License before Pavlov placed the SDK into the public domain in December 2008. LZMA typically compresses more tightly than older algorithms such as bzip2 while decompressing at a comparable speed, and its dictionary size can be scaled from a few kilobytes for embedded uses up to several gigabytes for maximum compression.