低秩矩阵恢复代码(共3页).doc

《低秩矩阵恢复代码(共3页).doc》由会员分享，可在线阅读，更多相关《低秩矩阵恢复代码(共3页).doc（3页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、精选优质文档-倾情为你奉上function x_out= MSB(Aop,y_vec,s_x,lambda1,rnk,iter) % function x_out obj_func= MSB(Aop,y_vec,m n,lambda1,rnk) % This code solves the problem of recovering a low rank matrix from its% lower dimensional projections % Minimize |X|* (nuclear norm of Z)% Subject to A(X) = Y %formulated as an

2、 unconstarined nuclear norm minmization problem using Split bregman algorithm % Minmimize (lambda1)|W|* + 1/2 | A(X) - y |_22 + eta/2 | W-X-B1 |_22%W is auxillary variable and B1 is the bregman variable %INPUTS%Aop : Linear operator%y_vec : Vector of observed values%s_x : size of the data matrix to

3、be recovered (in form of m n)%lambda1:regularization parameter%rnk : rank estimation for X%iter : maximum number of iterations to be carried out %OUTPUTS%x_out : recovered matrix if nargin 10 & abs(obj_func(iteration)-obj_func(iteration-10)1e-7 break end end %reshaping recovered vector to matrix for

4、m x_out = reshape(X,s_x(1),s_x(2); %Plot objective function plot(obj_func); title(Convergence of Objective Function) xlabel(Number of iterations); ylabel(Objective function value); function y = lsqrAop(x,transpose) switch transpose case transp y = Aop_concat(x,2); case notransp y = Aop_concat(x,1); end endend 专心-专注-专业