linear - Solving a underdetermined equation system in Eigen (C++) -

In my case, I want to solve an undated equation system, A * Î »= b, JacobiSVD

In the linear equation system, my C ++ program has the following structure:

Coefficient matrix ( A ): 0.6 5.68434e-20 -0.2 5.68434-20 7.06819-39 -4.26326e-20 -0.2 -4.26326e-20 0.4

RHS ( B ):

  -1.962 2.78817e-1 9 -5.886   
  Lumbda :  
  -9.81 3. 9 4467e + 19 & lt; ---------- Error (where does it come from?) -19.62   
 
 Rank of matrix A -  rank (a) = 2 < / Strong>  
 Then there is no complete space for the matrix. Therefore, A is singular and not inverse.  
 The condition is -  cond (A) = IP   
 To solve A * Î »= B, I will call it Eugene (JakobSVD)  SVD  used the decomposition method   
 I also verified it with MATLAB:  
 On the starting, the first simulation step is almost perfect but a lot There is a small numerical error, which is increasing while solving A * Î »= B.  
 Then the system is crashing and my results are no longer correct and I get NaN results.  
 Code here:  
  / ******** SVD ******** / Jacobi SVD & lt; TMatrixX & gt; SDD (A, ComputeneU | Computaneous); Lambda = svd.solve (b);   
 What have I done?   
 
 
 Jacobi SVD All non-resolved -zeros singular value I recommend using ColPivHouseholderQR .