[Octave-bug-tracker] [bug #30881] optim-1.0.12 nmsmax.m coding errors

[Robert Walton]

URL:
  <http://savannah.gnu.org/bugs/?30881>

                 Summary: optim-1.0.12 nmsmax.m coding errors
                 Project: GNU Octave
            Submitted by: bwalton
            Submitted on: Thu 26 Aug 2010 09:26:55 PM GMT
                Category: None
                Severity: 3 - Normal
              Item Group: None
                  Status: None
             Assigned to: None
         Originator Name: Robert Walton
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 3.2.4
        Operating System: Microsoft Windows

    _______________________________________________________

Details:

The optim-1.0.12 package's nmsmax.m (Nelder-Mead Simplex maximization)
contains three coding errors.  diff -c file fixing these problems is attached.
 The fixed file typically uses about 40% less function evaluations than the
original, and doesn't get "stuck" at non-optimum points anywhere near as often
(I don't want to claim "never").



    _______________________________________________________

File Attachments:


-------------------------------------------------------
Date: Thu 26 Aug 2010 09:26:55 PM GMT  Name: nmsmax_diff.txt  Size: 1kB   By:
bwalton
diff -c file to fix optim-1.0.12 nmsmax.m
<http://savannah.gnu.org/bugs/download.php?file_id=21317>

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?30881>

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