核范數(shù)和譜范數(shù)下廣義Sylvester方程最小二乘問題的有效算法
發(fā)布時(shí)間:2018-12-21 07:21
【摘要】:本文從數(shù)值角度討論Schatten q-范數(shù)下的廣義Sylvester方程約束最小二乘問題min x∈s‖N∑i=1A_iXB_i—C‖_q,其中S為閉凸約束集合,Schatten q-范數(shù)定義為‖M‖_q~q=∑_(i=1)~nσ_i~q(M),其中σ_i(M)為M∈R~(n×n)的奇異值.該問題的幾類特殊情形在圖像處理、控制論等領(lǐng)域有廣泛的應(yīng)用.q=2即Frobenius范數(shù)下該問題已被充分研究,故本文著重討論q=1,+∞,即核范數(shù)和譜范數(shù)下該問題的數(shù)值求解.采用的數(shù)值方法是非精確標(biāo)準(zhǔn)容易執(zhí)行的部分非精確交替方向法,并結(jié)合奇異值閾值算法,Moreau-Yosida正則化算法,譜投影算法和LSQR算法等求解相應(yīng)子問題.給出算法的收斂性證明,并用數(shù)值算例驗(yàn)證其高效可行性.
[Abstract]:In this paper, we discuss the constrained least squares problem of generalized Sylvester equation under Schatten q-norm from a numerical point of view, where S is a closed convex constrained set. The Schatten q-norm is defined as the singular value of M 鈭,
本文編號(hào):2388563
[Abstract]:In this paper, we discuss the constrained least squares problem of generalized Sylvester equation under Schatten q-norm from a numerical point of view, where S is a closed convex constrained set. The Schatten q-norm is defined as the singular value of M 鈭,
本文編號(hào):2388563
本文鏈接:http://sikaile.net/kejilunwen/yysx/2388563.html
最近更新
教材專著