An optimal algorithm to compute maximal relaxations in parallel

The study of constraint satisfaction problems touches many aspects of artificial intelligence. It is the premise of dominant interactive constraint satisfaction algorithms that users' preference has complete order. To accord more with practical situation, Haijiao Shen [3] did some research on the situation in which users' preferences have partial order and put forward the related algorithm in 2011. By introducing a constraint set M to decrease the spread of redundancy, we optimize her algorithm. We also prove the validity of our new algorithm and test it on some benchmarks. It is indicated by test result that the optimized algorithm MulExp can save up to 18% of examine times and up to 12% of solving time, which greatly increases the efficiency of solving process.