We consider the |$k$| mismatches version of approximate string matching for a single pattern and multiple patterns. For these problems, we present new algorithms utilizing the single instruction multiple data (SIMD) instruction set extensions for patterns of up to 32 characters. We apply SIMD computation in three ways: in counting of mismatches, in comparison of substrings and in calculation of fingerprints. We show the competitiveness of the new algorithms by practical experiments.

中文翻译：

---

SIMD近似字符串匹配

我们考虑| $ k $ | 单个模式和多个模式的近似字符串匹配版本不匹配。对于这些问题，我们提出了利用单指令多数据（SIMD）指令集扩展的新算法，最多可包含32个字符的模式。我们以三种方式应用SIMD计算：不匹配计数，子串比较和指纹计算。我们通过实际实验证明了新算法的竞争力。