[Octave-bug-tracker] [bug #55831] Incorrect sparse QR factorization

[anonymous]

URL:
  <https://savannah.gnu.org/bugs/?55831>

                 Summary: Incorrect sparse QR factorization 
                 Project: GNU Octave
            Submitted by: None
            Submitted on: Tue 05 Mar 2019 07:37:57 AM UTC
                Category: Octave Function
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Incorrect Result
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 5.1.0
        Operating System: Microsoft Windows

    _______________________________________________________

Details:

The following code:
[Q,R] = qr(sparse([1,0;0,0]));
outputs
Q = [-1 0
      0 0]

obviously Q is not unitary
it seems to me that each column with all zeros in a sparse input return a
column with all zeros in the Q matrix

the size of Q is also incorrect when the input is sparse and an economy size
factorization is requested. e.g.:
[Q,R] = qr(sparse(5,2),0);
outputs Q with size 5x5 rather than 5x2. Even though Q is 5x5, R is 2x2.

btw, is there a reason why Octave does not return a sparse Q like Matlab does
when the input is sparse? I know Q is usually dense, but in some cases it may
be sparse (e.g., economy factorization of a tall and sparse matrix). This
could be useful to avoid running out of memory, for instance.




    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?55831>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/