« Gauss partial pivoting » to « Gauss complete pivoting »

5 次查看（过去 30 天）

amine aquesbi 2020-1-20

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/501143-gauss-partial-pivoting-to-gauss-complete-pivoting

Hello, I want to make this gauss partial pivoting into a gauss complete pivoting but I can't fix it ... Can you help me with it please ?

% Gaussian inversion with pivoting 11/10/2019
function x=mygausspivot(A,b)
    if size(A,1) ~= size(A,2)
       error('Matrix is not square!)');
    end; 
    n=size(A,1);
        
    % Matrix triangualisation:
    for i=1:n
        % Searching for the pivoting line
        % disp([ 'Treatment of line i=' num2str(i) ] ); 
        n_max=i;
        for m=i:n
            if abs(A(n_max,i)) < abs(A(m,i))
                n_max=m;
            end
        end
	% disp([ 'Line with pivoting element n_max=' num2str(n_max) ] );
        % exchangement of line i and pivoting line n_max in the matrix A and in the vector b
        for l=i:n
            tmp=A(i,l);
            A(i,l)=A(n_max,l);
            A(n_max,l)=tmp;
        end
        tmp=b(i);
        b(i)=b(n_max);
        b(n_max)=tmp;
        p=A(i,i); % pivoting element
        % verification of non-generation of matrix
        if p==0
            disp('Pivoting element is zero (matrix could be degenerated)!!!');
            % error('Pivoting element is zero (matrix degenerated?)');
        end       
        for k=i:n
            A(i,k)=A(i,k)/p;
        end
        b(i)=b(i)/p;        
        for j=i+1:n
            r=A(j,i);
            for k=i:n
                A(j,k)=A(j,k)-A(i,k)*r;
            end
            b(j)=b(j)-b(i)*r;
        end
    end
    % In this point our matrix is, normally, upper triangular 
    % (with ones on the diaganonal), 
    % and we solving the folowing liniar equation:
    %  1    a_12 ...  a_1N      x_1      b_1
    %  0    1    ...  a_2N   *  x_2  =   b_2
    %  0    0    .........       .        .
    %  0    0    ...  1         x_N      b_N
    
    % x calculation from triangular matrix:   
    x=b;
    for i=n-1:-1:1
        for k=i+1:n
            x(i)=x(i)-x(k)*A(i,k);
        end
    end
     % now x containes the solution
     
end