【作者】 张娇娇； 丛爽； 郑凯； 李克之；

【Author】 Zhang Jiao-jiao;Cong Shuang;Zheng Kai;Li Ke-zhi;Automation Department of USTC;Imperial College London;

【机构】 中国科学技术大学自动化系； 英国帝国理工学院；

【摘要】 结合对偶上升法和扩展拉格朗日乘子法(ALM)的优点形成交替方向乘子法(Alternating Direction Multiplier Method,ADMM),本文提出了一种结合不动点方程的改进ADMM算法(简称FP ADMM)。FP ADMM是通过近邻算子获得最优解满足的不动点方程,再采用迭代法求解,避免了大规模矩阵求逆运算,降低了算法对计算机存储空间的要求,减少了在密度矩阵重构应用中的计算时间,使得高量子位系统的状态估计变得实际可行。仿真实验完成了量子比特位为8的量子系统密度矩阵估计,验证了所提出改进算法的优越性。更多还原

【Abstract】 The combination of the advantages of dual ascent algorithm and Augmented Lagrange Multiplier method(ALM) forms the Alternating Direction Multiplier Method(ADMM).This paper proposes an improved ADMM with fixed-point equation(called FP-ADMM).The proposed FP-ADMM can obtain the optimal solution satisfied the fixed-point equation with proximity operator,and the solution can be obtained by iterative method,which avoids solving the inversion of large scale matrix,reduces the need of storage space requirements of the computer and the computing time in density matrix reconstruction.In such a way,the estimation of quantum state with higher qubits becomes possible,The simulation experiments of estimating the quantum state in quantum system with 8-qubit are completed,which demonstrates the superiority of the improved algorithm proposed.更多还原