A Parallel Computation for Shift-Add Algorithm

The approximate string matching is useful in a wide area of applications such as biology. A practically significant speedup for solving this problem is obtained by representing strings as bit sequences and computing the comparisons of plural characters simultaneously by bit operations. In this method, a practical run-time depends on the word size of a computer. In this paper, as another parameter of the performance of a computer, the number of processors is considered. An efficient algorithm based on the previous speedup method is modified into a parallel algorithm. In the concrete, an $ O(mnlogm/w) $ algorithm for a problem of approximate string matching is modified to an $ O(mn/w)$ algorithm for a computer with m processors