MathGroup Archive 2001

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Re: A bug of Integrate[] in Mathematica 4.1 (and 4.0)

  • To: mathgroup at smc.vnet.net
  • Subject: [mg27443] Re: [mg27412] Re: A bug of Integrate[] in Mathematica 4.1 (and 4.0)
  • From: George Woodrow <ardaghconsultants at worldnet.att.net>
  • Date: Sun, 25 Feb 2001 20:55:47 -0500 (EST)
  • References: <972f8t$4ea@smc.vnet.net> <974vgh$6h1@smc.vnet.net> <200102250553.AAA10789@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

The bug is still there is 4.1. the integral is i/2 what it should be. 
Curiously, if (r-x) (r+x) is re-written as (r^2 - x^2), Mathematica gets the 
right answer. NIntegrate works correctly.

George Woodrow III


>Jens-Peer Kuska wrote:
>
>>  a) everyone that use a computer algebra and blind trust it's results
>>     is a born fool
>
>I might be a born fool but I'm never trusting CA systems in mathematical
>results ;-).
>
>>  b) don't call it "bug", call it feature
>
>Please call a bug a bug. I had a nice private discussion with someone
>from Wolfram Res. who admitted now that it is a bug. I'm aware of the
>fact that a fix of bugs of this kind is a rather complicated stuff.
>
>>  c) WRI cares about this features
>
>Well, I was a little too naughty. I wanted also to say that I find
>Mathematica a great tool, and it is impossible to get a multi purpose
>CA-system without bugs. I was only complaining about some support team
>emplyees who are telling me a bug is not a bug.
>
>>  d) a modification in Integrate[] needs probably a full new kernel
>
>Fine. How about version 4.1? Could someone check the my simple example
>
>Integrate[sqrt[(r-x)(r+x)],{x,-r,r}]
>
>
>>  g) when a sufficient number of features are corrected we get a version
>>     x.y.1  typical for free -- just a e-mail to the sales department is
>>     needed,  some days later a CD-ROM arrives you - can they do more ?
>
>Ok, is there such a CD-ROM for the student version of version 4.0?
>
>>  h) I assume that a public list of known bugs would be bad publicity
>>     because the most people believe that a software must be complete
>>     bug free to work at all. Bad sales due to confused customers
>>     ("What ? Mathematica has bugs ???" =:-O ) can't be in the interest
>>     of any Mathematica user.
>
>No, I think that the users of Mathematica are smart enough to like such
>bug reports. It is only helpful to avoid trouble from the very
>beginning. At the long sight it is a bad politics for a software company
>to hide errors. Wolfram research has some bug lists, as I have learned
>from my correspondence. It is not of my interest to make Mathematica
>worse saled than now.
>
>The behaviour you advertise for software companies might be the reason
>that sometimes non-commercial software is more stable than commercial.
>It is also very easy to communicate with open source software developers
>because they have no interest in hiding bugs but in finding bugs and
>cure them, and the best thing for a CA-system is to have bug reports
>from real users of the system, applying it in their practical work.
>Since Mathematica is so general there are a lot of different communities
>using it for a lot of different task so that it is easy to find many
>different kinds of bugs which again is a chance to fix them for the next
>version.
>
>But if I hear mathematically questionable arguments that a bug is not a
>bug or, even worse, that a bug is a feature as you did above, I have not
>much confidence that the bug will be fixed in a later version, because
>arguing in this way prevents developing better algorithms.
>
>
>--
>Hendrik van Hees		Phone:  ++49 6159 71-2751
>c/o GSI-Darmstadt SB3 3.183	Fax:    ++49 6159 71-2990
>Planckstr. 1			mailto:h.vanhees at gsi.de
>D-64291 Darmstadt		http://theory.gsi.de/~vanhees/index.html



  • Prev by Date: Re: Boundary value problem with NDSolve
  • Next by Date: "just a bug" !; Mathematica 4
  • Previous by thread: Re: A bug of Integrate[] in Mathematica 4.1 (and 4.0)
  • Next by thread: Re: A bug of Integrate[] in Mathematica 4.1 (and 4.0)