MWARRAY Matrix with multp array result return C#

Function Matla -- function [tt,pp,pr] = PCA(X,comp)

where input: X = Matrix; comp int;

and output is

tt = Matrix pp = Matrix pr = Matrix

So I'm not have know use tis dll of the PCA in program C# because have three different Matriz output.

   PCAdll.PCAClass test = new PCAdll.PCAClass(); // PCAdll is Dll creat in Matlab

        MWNumericArray MatrixInput= null;
        MWArray MatrixOut = null;
         int comp = 3;    

        matriz1 = new MWNumericArray(new double[10, 20]);

              for (int i = 1; i<5; i++)
               for(int j = 1; j<20; j++)
                 {
                     MatrixInput[i,j] = j;
                 }

         MatrixOut = test.PCA((MWArray) MatrixInput, comp);

so how to make to use it in C# program

---------------------- Function Funcion Matla -- function [tt,pp,pr] = PCA(X,comp)

function [tt,pp,pr] = PCA(X,comp)
% [tt,pp,pr]=PCA(X,comp);
% comp is the number of principal components
% Here the NIPALS algorithm is used
% pr = percentage explained variance 

X0 = X;
m = mean(X);
[a,b] = size(X);
X =  X - ones(a,1)*m;
% Centering the matrix
% the first approximation to t is the first
% column in X

stop = 3;
for i = 1:comp
i
av = std(X);

[av2,indx]=sort(av);
    t0 = X(:,indx(b));
       while stop > 1,
          p0 = t0'*X;
          p1 = p0/norm(p0);
          t1 = X*p1';
          if norm(t1-t0) < 0.00005,
             tt(:,i) = t1;
             pp(i,:) = p1;
             stop = 0;
          end;
          t0 = t1;
       end;
     stop = 3;
     X = X - t1*p1;
  end;

pr = (explv(X0,tt,pp))';

%eig = diag(tt'*tt);
%ssum = cumsum(eig);
%pr   = 100*ssum/ssum(comp);