MuPAD bugs: series can return 6 terms only
Vladimir Bondarenko
"Vladimir Bondarenko" <v...@cybertester.com> writes on
Tues, Feb 22 2005 9:30 pm
http://groups-beta.google.com/group/sci.math.symbolic/msg/e7d1c3f6cd9ec431
VB> You are a human (just like me). A weak, fallible creature
VB> with too imperfect memory of ridiculous size, painfully
VB> slow speed, very limited experience, and on top of it,
VB> absolutely unscalable (!)
VB> Just like me, you are absolutely ill-fitted for software
VB> testing.
Let us demonstrate you yet another MuPAD bug GEMM identified,
but a human tester failed to identify, just check out the
official MuPAD bug database at
http://research.mupad.de/bugs.shtml
MuPAD 3.1.1, MuPAD 3.1, MuPAD 3.0, MuPAD 2.5.2 can return
only 6 terms here.
> series(sech(z), z= infinity, 19);
2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 -
2/exp(z)^11 + O(1/exp(z)^13)
> series(sech(z), z= infinity, 199);
2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 -
2/exp(z)^11 + O(1/exp(z)^13)
> series(sech(z), z= infinity, 1999);
2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 -
2/exp(z)^11 + O(1/exp(z)^13)
Compare with the results from asympt. The results from these
two commands
series(sech(z), z= infinity, 19);
asympt(2/(exp(z)+1/exp(z)), z, 19);
must be identical but they are not.
> asympt(2/(exp(z)+1/exp(z)), z, 9);
2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 +
O(1/exp(z)^10)
> asympt(2/(exp(z)+1/exp(z)), z, 19);
2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 -
2/exp(z)^11 + 2/exp(z)^13 - 2/exp(z)^15 + 2/exp(z)^17 - 2/exp
(z)^19 + O(1/exp(z)^21)
Also compare with
> series(arctan(z), z= infinity, 1);
1/2*PI - 1/z + O(1/z^2)
> series(arctan(z), z= infinity, 5);
1/2*PI - 1/z + 1/3/z^3 - 1/5/z^5 + O(1/z^6)
> series(arctan(z), z= infinity, 10);
1/2*PI - 1/z + 1/3/z^3 - 1/5/z^5 + 1/7/z^7 - 1/9/z^9 + O(1/z^11)
> series(arctan(z), z= infinity, 20);
1/2*PI - 1/z + 1/3/z^3 - 1/5/z^5 + 1/7/z^7 - 1/9/z^9 + 1/11/z^11
- 1/13/z^13 + 1/15/z^15 - 1/17/z^17 + 1/19/z^19 + O(1/z^21)
The same bug manifestation with
series(csch(z), z= infinity, 19);
Best wishes,
VM and GEMM architect
Co-founder, CEO, Mathematical Director
Cyber Tester, LLC
http://www.cybertester.com/
http://maple.bug-list.org/
http://www.CAS-testing.org/
Will Twentyman
[bug report number who knows what]
Serious question: why are you posting these here instead of reporting
them to the people in charge of the MuPAD project?
--
Will Twentyman
email: wtwentyman at copper dot net
Julian Stoev
> Vladimir Bondarenko wrote:
>
> [bug report number who knows what]
>
> Serious question: why are you posting these here instead of reporting
> them to the people in charge of the MuPAD project?
I think it is great that he is posting these. He is doing some *real*
work to improve the quality of this software. I must say, that his posts
during the last one year made me suspusios about Maple results. The
reason why reporting to the sotware vendor is not sufficient is, because
it is difficult to find bug databases of different companies. I am not
sure it is even possible to get to the bug data for Mathematica and
Maple from outside.
Maybe he should post only in sci.math.symbolic.
--JS
Patrick Hamlyn
>Vladimir Bondarenko wrote:
>
>[bug report number who knows what]
>
>Serious question: why are you posting these here instead of reporting
>them to the people in charge of the MuPAD project?
Kinda obvious isn't it.... he's trying to push an agenda. 'Damage the reputation
of various computer algebra systems publically' rather than 'improve various
CASs by reporting bugs'.
The real question is why he's bothering...
--
Patrick Hamlyn posting from Perth, Western Australia
Windsurfing capital of the Southern Hemisphere
Moderator: polyforms group (polyforms...@egroups.com)
Brad Cooper
Good to see an Aussie from the other side of the continent having 2 bob's
worth in this newsgroup :-)
> Will Twentyman <wtwen...@read.my.sig> wrote:
>
> >Vladimir Bondarenko wrote:
> >
> >[bug report number who knows what]
> >
> >Serious question: why are you posting these here instead of reporting
> >them to the people in charge of the MuPAD project?
>
> Kinda obvious isn't it.... he's trying to push an agenda. 'Damage the
reputation
> of various computer algebra systems publically' rather than 'improve
various
> CASs by reporting bugs'.
I appreciate the reports in this newsgroup from the GEMM project for two
reasons:
1. I feel I am gaining an insight into the relative quality and strengths of
the various CASs on the market.
2. You say that Vladimir should 'improve various CASs by reporting bugs',
but not publicly. The GEMM project shows to me that one company in
particular does not fix their bugs when they are reported publicly or
otherwise. Without the work of Vladimir Bondarenko I would be blissfully
unaware of the lack of regression testing with this CAS. I almost bought it
at one stage.
I remember going to a presentation by the team leader for FORTRAN compiler
development at Digital Equipment Corporation. I was particularly impressed
with their regression testing. He said that every new release of the
compiler, without exception, had to correctly compile every piece of code
that had produced a problem in any release since version 1.0
I recall an anecdote in which the speaker said a neighbour of his was not a
professional programmer, but he had produced some spaghetti FORTRAN to
maintain the results ladder for his ten pin bowling team. The code didn't
work because a weird construct in the user's program actually exercised a
bug in the compiler. The bug was fixed and even that piece of code for the
bowling club was included in the regression testing for all further releases
of the compiler.
I don't want to go on for too long here, but I would like to round out the
story. At that conference the very next speaker was the woman in charge of
development of sofware for the flight controls of the new (at that time)
Boeing 737. It was written using the FORTRAN compiler from Digital Equipment
Corporation. She said that it was chosen because of the extensive regression
testing of the compiler and its acknowledged reputation for quality in the
industry which was verified by their own testing. That gave me a warm and
fuzzy feeling when I flew in the first commercial flights of the 737.
The postings of Vladimir Bondarenko, in this newsgroup, clearly demonstrate
to me the degree to which various CAS producers have that same committment
to quality. If the same bugs exist through numerous software releases, I
think that is valuable public information. It just should not happen.
Cheers,
Brad
Richard Kanarek
I can't say that I agree with your (rather harsh) assessment of Mr.
Bondarenko.
Although I'm sure he can state his reasons better than I can, it would
appear that Mr. Bondarenko's efforts are "motivated" by a particular
dislike of CAS software defects ("bugs") and a particular fondness for
the automated method he developed (and is marketing?) for detecting
defects.
I think that castigating Mr. Bondarenko for not "reporting" bugs is
wholly incorrect. I don't see how Mr. Bondarenko, who is presumably
not employed by one of the CAS producers, can be considered obligated
to report their bugs in strict accordance with their rules. Further,
posting reports of the bugs here ought to be enough. If a little (?)
company/group like Sciface, or at least an occasional
employee/associate/(indentured grad student <g>) can bother to stop by
every now and again, surely someone from the other companies could be
bothered to too.
Lastly, I'm glad that Mr. Bondarenko's loose cannon has swung around
to MuPAD. I'm currently a reasonably "happy" MuPad user/Sciface
customer. If Mr. Bondarenko can help keep SciFace focused on quality
improvements, while warning me of current problems, I can only get
happier.
Stefan Wehmeier
> Vladimir Bondarenko wrote:
>
> [bug report number who knows what]
>
> Serious question: why are you posting these here instead of reporting
> them to the people in charge of the MuPAD project?
>
Never mind - many of us at MuPAD read this group regularly. Actually,
Vladimir has contributed many bugs to
www.mupad.de/BUGS
and even got an award for being our best beta tester.
The advantage of using our bug interface is that you get a status report
whenever something changes (e.g. the bug is fixed, or a developer gives a
detailed analysis why something is going wrong). If people write about us
(and not to us), we will take notice but not answer in all cases.
What I do not understand is what all these bugs have to do with Maple or
with recreational mathematics. Therefore, this message goes to
sci.math.symbolic only.
--
Stefan Wehmeier
ste...@math.uni-paderborn.de
Alec Mihailovs
"Stefan Wehmeier" <ste...@mupad.de> wrote in message
news:cvkb1d$ckt$1...@news.uni-paderborn.de...
> Actually,
> Vladimir has contributed many bugs to
> www.mupad.de/BUGS
> and even got an award for being our best beta tester.
You mean a T-shirt?
Will Twentyman
> Greetings,
>
> I can't say that I agree with your (rather harsh) assessment of Mr.
> Bondarenko.
>
> Although I'm sure he can state his reasons better than I can, it would
> appear that Mr. Bondarenko's efforts are "motivated" by a particular
> dislike of CAS software defects ("bugs") and a particular fondness for
> the automated method he developed (and is marketing?) for detecting
> defects.
>
> I think that castigating Mr. Bondarenko for not "reporting" bugs is
> wholly incorrect. I don't see how Mr. Bondarenko, who is presumably
> not employed by one of the CAS producers, can be considered obligated
> to report their bugs in strict accordance with their rules. Further,
> posting reports of the bugs here ought to be enough. If a little (?)
> company/group like Sciface, or at least an occasional
> employee/associate/(indentured grad student <g>) can bother to stop by
> every now and again, surely someone from the other companies could be
> bothered to too.
Here's the problem: Is there any evidence that anyone from those groups
is actually seeing these? The people who read these groups are a
self-selecting group, who may not be in a position to openly discuss
their products or research. I happened to learn about usenet in the
early 90s while I was in college. I was away from it for a while, and
have since returned to reading it. Unfortunately, in the current
society we have, this medium is also viewed by some as something of a
dinosaur. To expect the people working on these systems to be here is
unreasonable. I hope a few are reading, but I wouldn't bet on it.
> Lastly, I'm glad that Mr. Bondarenko's loose cannon has swung around
> to MuPAD. I'm currently a reasonably "happy" MuPad user/Sciface
> customer. If Mr. Bondarenko can help keep SciFace focused on quality
> improvements, while warning me of current problems, I can only get
> happier.
Thus my question. Why not directly communicate with those who can fix
the software, instead of broadcasting the findings to us? They may not
fix it, but at least then he could be certain it was known.
Brad Cooper
"Brad Cooper" <Brad.Co...@bigpond.com> wrote in message
news:43bTd.173048$K7.7...@news-server.bigpond.net.au...
[snip]
> one company in
> particular does not fix their bugs when they are reported publicly or
> otherwise. Without the work of Vladimir Bondarenko I would be blissfully
> unaware of the lack of regression testing with this CAS. I almost bought
it
> at one stage.
Because the subject of the message to which I was replying was MuPad and I
did not make specify the name of the company or product with the poor record
on regression testing, I may have given the impression my disappointment is
with SciFace/MuPad. In fact that is not the case. I have actually bought
MuPad and am generally very happy with it.
None of the CASs are perfect, but Maple would appear to be the CAS which
suffers most from lack of regression testing - according to the GEMM
project. I felt that I should clarify what I posted because a direct email
response to me showed I had created the wrong impression, with one person at
least.
Cheers,
Brad
Christopher Creutzig
> during the last one year made me suspusios about Maple results. The
> reason why reporting to the sotware vendor is not sufficient is, because
> it is difficult to find bug databases of different companies. I am not
It should be easy enough to find http://research.mupad.de/bugs.html.
I don't know about the other systems, though.
> Maybe he should post only in sci.math.symbolic.
Personally, I agree. Excessive cross-posting has never been a good idea.
rgeards,
Christopher Creutzig
Christopher Creutzig
>>series(sech(z), z= infinity, 199);
>
>
> 2/exp(z) - 2/exp(z)^3 + 2/exp(z)^5 - 2/exp(z)^7 + 2/exp(z)^9 -
> 2/exp(z)^11 + O(1/exp(z)^13)
In case anyone is looking for a workaround: Setting ORDER:=199 instead
of supplying the third argument to series does help. Note that for
generalized series as the one presented here, the notion of "ORDER" is
(necessarily, in general) a bit fuzzy; in fact, with ORDER=40, I get a
series expansion with 79 terms.
regards,
Christopher Creutzig