[Octave-bug-tracker] [bug #32502] cov function returns incorrect results for complex matrices

[Ivan]

Follow-up Comment #2, bug #32502 (project octave):

Here goes octave:

 A_r = load("-ascii", "real_input.ascii");
 A_i = load("-ascii", "imag_input.ascii");
 A = complex(A_r, A_i);
 imag(A)

ans =

  -0.00000  -0.25112   0.16022  -0.44749  -0.06819   0.23258  -0.05861 
-0.06876
   0.25112  -0.00000  -0.07418  -0.05551   0.11349   0.01903   0.15368 
-0.09032
  -0.16022   0.07418  -0.00000   0.28853  -0.07469  -0.16844   0.18365  
0.02444
   0.44749   0.05551  -0.28853  -0.00000   0.12424   0.04581  -0.23836  
0.10005
   0.06819  -0.11349   0.07469  -0.12424  -0.00000   0.03534  -0.20799  
0.32262
  -0.23258  -0.01903   0.16844  -0.04581  -0.03534  -0.00000  -0.34949  
0.11833
   0.05861  -0.15368  -0.18365   0.23836   0.20799   0.34949  -0.00000  
0.15882
   0.06876   0.09032  -0.02444  -0.10005  -0.32262  -0.11833  -0.15882 
-0.00000


Since matlab is prohibited (and actually I don't have access to it right now),
a C++ code snippet, doing the same (it uses armadillo
<http://arma.sourceforge.net/> library):

#include <armadillo>
#include <iostream>

int main() {
    using namespace arma;
    mat real_input, imag_input;
    real_input.load("data/real_input.ascii", raw_ascii);
    imag_input.load("data/imag_input.ascii", raw_ascii);
    cx_mat input(real_input, imag_input);
    std::cout << imag(cov(input)) << "n";
}

$ g++ test.cc -I../../../contrib -llapack -lblas
$ ./a.out
        0   0.2511  -0.1602   0.4475   0.0682  -0.2326   0.0586   0.0688
  -0.2511        0   0.0742   0.0555  -0.1135  -0.0190  -0.1537   0.0903
   0.1602  -0.0742        0  -0.2885   0.0747   0.1684  -0.1837  -0.0244
  -0.4475  -0.0555   0.2885        0  -0.1242  -0.0458   0.2384  -0.1000
  -0.0682   0.1135  -0.0747   0.1242        0  -0.0353   0.2080  -0.3226
   0.2326   0.0190  -0.1684   0.0458   0.0353        0   0.3495  -0.1183
  -0.0586   0.1537   0.1837  -0.2384  -0.2080  -0.3495        0  -0.1588
  -0.0688  -0.0903   0.0244   0.1000   0.3226   0.1183   0.1588        0

As you can see this matrix are complex conjugates of each other. Matlab yield
the same result, btw.

(file #22704, file #22705)
    _______________________________________________________

Additional Item Attachment:

File name: real_input.ascii               Size:1 KB
File name: imag_input.ascii               Size:1 KB


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?32502>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/