实对称半正定矩阵恢复的Lagrange乘子修正算法

基于不精确的增广拉格朗日乘子算法,针对实对称半正定矩阵恢复问题提出了一种修正算法.恢复后的矩阵保持稳定的实对称半正定性质.同时,证明了修正算法的收敛性,验证了修正算法对实对称半正定矩阵恢复具有更高的效率.

A modified method based on the augmented Lagrange multiplier method for the real symmetric positive semidefinite matrix recovery

This paper proposes a modified method for the real symmetric positive semidefinite matrix recovery based on the Inexact Augmented Lagrange Multiplier Method. The recovered matrix keeps a feasible symmetric positive semidefinite structure. Meanwhile,it proves the convergence of the modified algorithm. Finally,it shows the modified algorithm is much more effective than the Inexact Augmented Lagrange Multiplier algorithm under some reasonable conditions through numerical experiments.