作者: Kimmo Fredriksson , Szymon Grabowski
DOI: 10.1016/J.EJC.2012.07.013
关键词: Mathematics 、 Combinatorics 、 Word ram 、 Binary logarithm 、 Approximate string matching 、 Word (computer architecture) 、 Parallelism (rhetoric) 、 Algorithm 、 Model of computation 、 Hamming distance
摘要: We develop a method for performing convolutions efficiently in word RAM model of computation, having size w = ? ( log n ) bits, where is the input size. The basic?idea?is to pack several elements vector into single computer word, effectively enabling parallel computation convolutions. technique applied approximate string matching under Hamming distance. obtained algorithms are fastest known. In particular, we reduce complexity Amir et?al.?(2000) algorithm k -mismatches from O + / . Those impose some (not severe) limitation on pattern length, m present another, less efficient however, based word-level parallelism, which works without length limitation.