摘要
讨论信号恢复问题,对l_(1)正则化模型,用光滑函数近似l_(1)-范数,并用三项共轭梯度法进行求解。证明了水平集的有界性,函数梯度的Lipschitz连续性,得到了算法的全局收敛性。进行了数值实验,数值实验结果表明本文方法的有效性。
The signal recovery problem is discussed.For the l_(1)-norm regularization model,the smoothing function is used to approximate the l_(1)-norm,then a three-terms conjugate gradient algorithm is used to solve it.The boundedness of the level set and the Lipschitz continuity of the target function’s gradient are proved.The global convergence of our algorithm is obtained.The numerical results show that our method is effective.
作者
王静
乌彩英
WANG Jing;WU Caiying(School of Mathematical Science,Inner Mongolia University,Hohhot 0100219,China)
出处
《内蒙古大学学报(自然科学版)》
CAS
北大核心
2021年第1期27-32,共6页
Journal of Inner Mongolia University:Natural Science Edition
基金
内蒙古自然科学基金(2018MS01016)。