octave-bug-tracker
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Octave-bug-tracker] [bug #40895] Being less strict about three test fai


From: Pascal Dupuis
Subject: [Octave-bug-tracker] [bug #40895] Being less strict about three test failures in conv2.cc ?
Date: Sat, 14 Dec 2013 11:56:14 +0000
User-agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/31.0.1650.63 Safari/537.36

URL:
  <http://savannah.gnu.org/bugs/?40895>

                 Summary: Being less strict about three test failures in
conv2.cc ?
                 Project: GNU Octave
            Submitted by: cdemills
            Submitted on: sam. 14 déc. 2013 11:56:13 GMT
                Category: Libraries
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Inaccurate Result
                  Status: None
             Assigned to: None
         Originator Name: Pascal Dupuis
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: dev
        Operating System: GNU/Linux

    _______________________________________________________

Details:

Hello,

while looking at the known test failures, I found three occurences in
conv2.cc. The comments state that there are two computation paths, with
different ordering of multiplications and summation. Truncation errors
accumulate in the order of tens of epsilon. I propose a patch to make the
disregard this truncation error and remove this known failure. 

Patch included

Regards

Pascal



    _______________________________________________________

File Attachments:


-------------------------------------------------------
Date: sam. 14 déc. 2013 11:56:13 GMT  Name: conv2.cc.diff  Size: 2 ko   By:
cdemills

<http://savannah.gnu.org/bugs/download.php?file_id=29919>

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?40895>

_______________________________________________
  Message posté via/par Savannah
  http://savannah.gnu.org/




reply via email to

[Prev in Thread] Current Thread [Next in Thread]