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

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

Re: [Octave-bug-tracker] [bug #59807] quadgk results of v.6 different fr


From: Vougas D . V .
Subject: Re: [Octave-bug-tracker] [bug #59807] quadgk results of v.6 different from results of v.5 (correct ones as checked with MATLAB)
Date: Tue, 05 Jan 2021 06:34:30 -0000

I have not used Waypoints with quadgk, I have used the defaults only.  I believe there are still loose ends with new quadgk. The old v.5 version works fine for my problem. However quadgk in v.6 is problematic. In a machine with v.6 only installed it gives incorrect results. In a machine with both v.5 and v.6 installed, quadgk in v.6 gives correct results but the display of results is with 4 instead of 5 digits. Why quadgk gives incorrect results if v.5 is not already installed along v.6? Why is the default display in v.6 reduced to 4 digits? These are the problems. 

PS. The integral/quadcc routine of Octave is horrible.
   


From: Rik <INVALID.NOREPLY@gnu.org>
Sent: 04 January 2021 16:40
To: Markus Mützel <markus.muetzel@gmx.de>; d.v.vougas@swan.ac.uk <d.v.vougas@swan.ac.uk>; octave-bug-tracker@gnu.org <octave-bug-tracker@gnu.org>
Subject: [bug #59807] quadgk results of v.6 different from results of v.5 (correct ones as checked with MATLAB)
 
Update of bug #59807 (project octave):

                 Summary: Results of v.6 different from results  of v.5
(correct ones as checked with MATLAB) => quadgk results of v.6 different from
results  of v.5 (correct ones as checked with MATLAB)

    _______________________________________________________

Follow-up Comment #2:

>From the NEWS file for the 6.1 release


- Numerical integration has been improved.  The `quadv` function has
  been re-written so that it can compute integrands of periodic
  functions.  At the same time, performance is better with ~3.5X fewer
  function evaluations required.  A bug in `quadgk` that caused complex
  path integrals specified with `"Waypoints"` to occasionally be
  calculated in the opposite direction was fixed.


We need some examples to ascertain whether the *differences* between v5 and v6
might be related to the use of "Waypoints".

    _______________________________________________________

Reply to this item at:

  <https://eur03.safelinks.protection.outlook.com/?url="">>

_______________________________________________
  Message sent via Savannah
 
https://eur03.safelinks.protection.outlook.com/?url="">


reply via email to

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