The Research And Design Of Quantum Algorithm And Its Application

Quantum information is the combination during the development of quantum mechanicsand information science. It is of significance to research on quantum computing due to itshigh speed and high effect in theory. Because the specialty of quantum algorithm showsincomparable superiority in some fields, more quantum algorithms are designed and appliedto deal with traditional problems which classical computer can’t resolve.The research of this article is mainly based on existing quantum algorithms and improvesthem to be well done in some specific problems. Primary work can be concluded as follows:(1) Modify Grover’s algorithm, and propose a model for quantum pattern searchGrover’s algorithm is a quantum algorithm based on oracle. Boyer et al. then presented aframework named BBHT after their deep analysis and prove the effectiveness of Grover’salgorithm. In this article, a novel quantum initial state is designed to achieve quantum patternstorage. Using the BBHT as the reference and one kind of traditional pattern matchmechanism helps to quantum pattern search successfully. When queried pattern is not exactlymatch with any pattern stored in the memory, algorithm can give a close solution(2) Propose Bloch based quantum multi-objective evolutionary algorithm–BQMOEAEvolutionary Algorithm (EA) is a kind of heuristic optimization algorithm, andmulti-objective evolutionary algorithm can be used to work out multi-objective problems.Quantum multi-objective evolutionary algorithm (QMOEA) is more effective because thecombination of quantum concept and traditional algorithm. It describes that population isencoded by Bloch coordinates, and the population evolution utilizes quantum rotation gate inthis article. The kernel part makes several NSGA-II and improves it so that it can bringdiversity and uniformly distribution to the Pareto front. And some benchmark functions areused to test the effectiveness and robustness of the algorithm. MATLAB program simulatesBQMOEA and gives the testing results in this article.(3) Optimize MSCC using proposed BQMOEAClustering and classification which are the common problem in pattern recognition arealso known as unsupervised pattern classification and supervised pattern classificationrespectively. Multi-objective simultaneous learning framework for clustering andclassification (MSCC) was first proposed in2010and it was optimized by multi-objectiveparticle swarm algorithm (MOPSO). But it was obvious that few Pareto solutions againstpopulation optimization. In this article, BQMOEA replaces MOPSO to optimize MSCC, itcan get more Pareto solutions so that a higher classification accuracy rate. And artificialdataset and real datasets are both used to verify and test.