<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Will,<br>
    <br>
    I tried your suggestion, but all I got back was missing values.<br>
    <br>
    After Dennis's suggestion, I tried replacing the zeros with tiny
    random numbers, using this code:<br>
    <blockquote><tt>R = random_normal( 0.,1e-12, (/nlev*2,nlev*2/))</tt><tt><br>
      </tt><tt>M = where(M.eq.0,R,M)</tt><tt><br>
      </tt><tt>iM = inverse_matrix(M)</tt><br>
    </blockquote>
    Interestingly, this worked perfectly, and I got what I expected.<br>
    I also tried using this approach to trying Will's suggestion, and
    that also worked with the random numbers!<br>
    <br>
    I can't figure out a mathematical explanation, but this seems like a
    pretty nice fix. Might be good to add this to the online
    documentation.<br>
    <br>
    Thanks,<br>
    Walter<br>
    <br>
    <div class="moz-cite-prefix">On 11/10/14 6:17 PM, Will Hobbs wrote:<br>
    </div>
    <blockquote cite="mid:D0879173.90CA%25will.hobbs@utas.edu.au"
      type="cite">
      <meta http-equiv="Content-Type" content="text/html;
        charset=windows-1252">
      <div>Walter</div>
      <div><br>
      </div>
      <div>Instead of using the inverse matrix() function, try using
        solve_linsys(A, B) instead, for A =  your matrix and B = I (the
        identity matrix). Since A * inv(A) = I, the output will be the
        inverse.</div>
      <div><br>
      </div>
      <div>Mathematically this is no different to inverse_matrix(), but
        the code is a little different. In particular, if the matrix is
        singular (i.e. the determinant = 0, and it cannot be inverted)
        then it will return an error code.</div>
      <div><a moz-do-not-send="true"
href="https://www.ncl.ucar.edu/Document/Functions/Built-in/solve_linsys.shtml">https://www.ncl.ucar.edu/Document/Functions/Built-in/solve_linsys.shtml</a></div>
      <div><br>
      </div>
      <div>This might help you figure out if the matrix itself is the
        problem (I suspect, as Dennis has suggested, that may be the
        issue).</div>
      <div><br>
      </div>
      <div>Will</div>
      <div><br>
      </div>
      <div><br>
      </div>
      <span id="OLK_SRC_BODY_SECTION">
        <div style="font-family:Calibri; font-size:11pt;
          text-align:left; color:black; BORDER-BOTTOM: medium none;
          BORDER-LEFT: medium none; PADDING-BOTTOM: 0in; PADDING-LEFT:
          0in; PADDING-RIGHT: 0in; BORDER-TOP: #b5c4df 1pt solid;
          BORDER-RIGHT: medium none; PADDING-TOP: 3pt">
          <span style="font-weight:bold">From: </span>Walter Hannah
          &lt;<a moz-do-not-send="true"
            href="mailto:whannah@atmos.colostate.edu">whannah@atmos.colostate.edu</a>&gt;<br>
          <span style="font-weight:bold">Date: </span>Tuesday, 11
          November 2014 9:52 AM<br>
          <span style="font-weight:bold">To: </span>Dennis Shea &lt;<a
            moz-do-not-send="true" href="mailto:shea@ucar.edu">shea@ucar.edu</a>&gt;<br>
          <span style="font-weight:bold">Cc: </span>"<a
            moz-do-not-send="true" href="mailto:ncl-talk@ucar.edu">ncl-talk@ucar.edu</a>"
          &lt;<a moz-do-not-send="true" href="mailto:ncl-talk@ucar.edu">ncl-talk@ucar.edu</a>&gt;<br>
          <span style="font-weight:bold">Subject: </span>Re: [ncl-talk]
          inverse_matrix problem<br>
        </div>
        <div><br>
        </div>
        <div>
          <div bgcolor="#FFFFFF" text="#000000">Dennis,<br>
            <br>
            Wow! Thanks for going above and beyond! This helps a lot.
            I'll try to look for a way to get a modified matrix that
            doesn't have all those zeros.<br>
            <br>
            Walter<br>
            <br>
            <div class="moz-cite-prefix">On 11/10/14 5:50 PM, Dennis
              Shea wrote:<br>
            </div>
            <blockquote
cite="mid:CAOF1d_7VTz2TU8BRDMqZFrD4r=w0_E+gbnXHWbUC2YNvQuA3hQ@mail.gmail.com"
              type="cite">
              <div dir="ltr">
                <div>
                  <div>
                    <div>You have rows of 0.0 ... I think that is
                      causing issues. I'm sure some 'math-whiz' could
                      answer but I attached a simple test code that
                      reads your data (thx for providing). It calculates
                      the inverse and the identity matrix. Note that NCL
                      is using a direct call to the documented LAPACK
                      routines.<br>
                      <br>
                      ===<br>
                    </div>
                    TEST1: duplicates your findings<br>
                    <br>
                  </div>
                  TEST 2: create a bogus 20x20 array of random numbers.
                  No row consists of all constants. Identity matrix is
                  successfully calculated.<br>
                  <br>
                </div>
                TEST 3: wherever the original array had a row of all
                0.0, substitute random numbers. The identity matrix is
                successfully calculated.</div>
              <div class="gmail_extra"><br>
                <div class="gmail_quote">On Mon, Nov 10, 2014 at 1:57
                  PM, Walter Hannah <span dir="ltr">
                    &lt;<a moz-do-not-send="true"
                      href="mailto:whannah@atmos.colostate.edu"
                      target="_blank">whannah@atmos.colostate.edu</a>&gt;</span>
                  wrote:<br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    <div bgcolor="#FFFFFF" text="#000000">Hi,<br>
                      <br>
                      I'm trying to use the "inverse_matrix" function.
                      The matrix doesn't have any missing values, but I
                      still get this error message:<br>
                      <blockquote><tt>Variable: M</tt><tt><br>
                        </tt><tt>Type: float</tt><tt><br>
                        </tt><tt>Total Size: 1600 bytes</tt><tt><br>
                        </tt><tt>            400 values</tt><tt><br>
                        </tt><tt>Number of Dimensions: 2</tt><tt><br>
                        </tt><tt>Dimensions and sizes:    [pertlev | 20]
                          x [tendlev | 20]</tt><tt><br>
                        </tt><tt>Coordinates:</tt><tt><br>
                        </tt><tt>            pertlev: [50..1950]</tt><tt><br>
                        </tt><tt>            tendlev: [50..1950]</tt><tt><br>
                        </tt><tt>Number Of Attributes: 2</tt><tt><br>
                        </tt><tt>  missing_value :    9.96921e+36</tt><tt><br>
                        </tt><tt>  _FillValue :    9.96921e+36</tt><tt><br>
                        </tt><tt>(0)   
----------------------------------------------------------------------------</tt><tt><br>
                        </tt><tt><br>
                        </tt><tt>   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00<br>
                             0.00  -0.31  -0.06   0.15  -0.27  -0.56 
                          -1.21  -1.38  -1.36   1.49   0.00   0.00  
                          4.65   0.37   0.46   0.47   0.52   0.55  
                          0.77   2.49<br>
                             0.00   0.26  -0.35  -0.29  -0.65  -1.13 
                          -2.24  -2.45  -2.47   2.91   0.00   0.00  
                          4.07   2.45   0.95   0.87   0.90   1.05  
                          1.55   4.81<br>
                             0.00  -0.11   1.43  -0.91  -1.69  -1.52 
                          -2.63  -2.81  -2.69   3.17   0.00   0.00 
                          -2.35   2.61   1.49   1.01   1.00   1.18  
                          1.78   5.32<br>
                             0.00  -0.02   0.55   1.08  -1.74  -2.04 
                          -2.59  -2.71  -2.39   2.63   0.00   0.00 
                          -0.75  -1.10   1.35   1.02   0.89   1.00  
                          1.57   4.87<br>
                             0.00  -0.06   0.16   0.52   0.72  -1.86 
                          -2.88  -2.74  -2.29   2.31   0.00   0.00  
                          0.31  -1.30  -0.26   0.87   0.75   0.82  
                          1.27   4.50<br>
                             0.00  -0.13  -0.07   0.20   0.54   0.27 
                          -2.74  -3.10  -2.27   2.04   0.00   0.00  
                          0.08  -0.49  -0.47   0.14   0.72   0.69  
                          1.13   4.24<br>
                             0.00  -0.15  -0.06   0.20  -0.17   0.02 
                          -0.77  -3.87  -3.13   2.39   0.00   0.00  
                          0.69  -0.08   0.09   0.03   0.36   0.90  
                          1.29   4.59<br>
                             0.00  -0.14   0.05   0.09  -0.17  -0.29 
                          -0.14  -0.66  -5.36   2.00   0.00   0.00  
                          0.39   0.17   0.14   0.07  -0.01   0.22  
                          1.00   4.05<br>
                             0.00   0.02  -0.07  -0.01   0.00   0.02  
                          0.43   0.94   0.77  -2.90   0.00   0.00 
                          -0.09   0.09  -0.04  -0.07  -0.17  -0.26 
                          -0.49   0.54<br>
                             0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00<br>
                             0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00   0.00   0.00   0.00   0.00  
                          0.00   0.00<br>
                             0.00  -0.01  -0.04   0.12  -0.02  -0.01 
                          -0.03  -0.05  -0.05   0.04   0.00   0.00 
                          -2.17  -0.25  -0.14   0.01   0.02   0.02  
                          0.02   0.12<br>
                             0.00   0.00  -0.19   0.11   0.28   0.06  
                          0.11   0.06   0.07  -0.13   0.00   0.00  
                          2.75  -1.97  -0.04  -0.12  -0.05  -0.05 
                          -0.08  -0.10<br>
                             0.00  -0.02  -0.14  -0.08   0.31   0.21  
                          0.13   0.08   0.06  -0.08   0.00   0.00  
                          0.48   1.11  -1.09  -0.13  -0.10  -0.06 
                          -0.09  -0.12<br>
                             0.00  -0.01  -0.03  -0.11  -0.07   0.32  
                          0.27   0.12   0.18  -0.17   0.00   0.00  
                          0.02   0.63   0.52  -0.94  -0.02  -0.15 
                          -0.08  -0.25<br>
                             0.00   0.02   0.08  -0.06  -0.24   0.10  
                          0.43   0.26   0.16  -0.17   0.00   0.00  
                          0.07   0.31   0.35   0.40  -1.06  -0.06 
                          -0.20  -0.26<br>
                             0.00   0.06   0.03  -0.10   0.07   0.11  
                          1.03   1.38   0.53  -1.10   0.00   0.00 
                          -0.44   0.14  -0.03   0.08   0.19  -1.61 
                          -0.76  -1.16<br>
                             0.00   0.11  -0.03  -0.08   0.07   0.16  
                          0.46   1.28   1.65  -1.21   0.00   0.00 
                          -0.25   0.07  -0.06   0.00   0.14   0.41 
                          -3.04  -0.66<br>
                             0.00   0.12  -0.30  -0.15   0.80   1.58  
                          2.86   3.62   5.02  -3.15   0.00   0.00 
                          -1.23  -0.91  -0.80  -0.94  -0.85  -0.66  
                          0.21 -10.64<br>
                          <br>
                          (0)   
----------------------------------------------------------------------------<br>
                          (0)    # missing values: 0<br>
                          (0)    # values = 0.0  : 111</tt><tt><br>
                        </tt><tt>warning:inverse_matrix: info = 1;
                          missing values not allowed</tt><br>
                      </blockquote>
                      <br>
                      I thought it might have something to do with those
                      zero values, but if I replace them with non-zero
                      values I get a similar, but not identical, result:<br>
                      <blockquote><tt>...<br>
                          (0)   
----------------------------------------------------------------------------</tt><tt><br>
                        </tt><tt>(0)    # missing values: 0</tt><tt><br>
                        </tt><tt>(0)    # values = 0.0  : 0</tt><tt><br>
                        </tt><tt>warning:inverse_matrix: info = 20;
                          missing values not allowed</tt><br>
                      </blockquote>
                      <br>
                      <br>
                      Thanks,<br>
                      Walter </div>
                    <br>
                    _______________________________________________<br>
                    ncl-talk mailing list<br>
                    List instructions, subscriber options, unsubscribe:<br>
                    <a moz-do-not-send="true"
                      href="http://mailman.ucar.edu/mailman/listinfo/ncl-talk"
                      target="_blank">http://mailman.ucar.edu/mailman/listinfo/ncl-talk</a><br>
                    <br>
                  </blockquote>
                </div>
                <br>
              </div>
            </blockquote>
            <br>
          </div>
        </div>
      </span>
    </blockquote>
    <br>
  </body>
</html>