Author 
Message 
Allan Adler science forum beginner
Joined: 25 Mar 2005
Posts: 30

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Steps towards writing a computer algebra system



Allan Adler <ara@nestle.csail.mit.edu> writes:
Quote:  I'm glad to learn of this development. I'm downloading the book and
the source now.

OK, I've downloaded both and installed Axiom, more of less. It seems to
have gobbled up about 500 MB of disk space. Make didn't run to completion.
I started it around 11pm last night and it was still running when I went
to sleep. I woke up once in the middle of the night and saw it in the
middle of something and when I finally woke up a few hours later, it was
still hung up at that same point. I stopped the Make and looked around
to see if I could simply run Axiom, which I could. Then I restarted Make
and quickly got to the point where it had been hung up, which turned out
to be a file called int/input/r21bugsbig.input, specifically the line:
g : LX := reduce(*, [X  rho for rho in r]) ;
At the beginning of the file, it says that it takes a long time, but I have
no way of figuring out whether this particular computation will ever halt.
I don't know how much memory it takes to run this, so the problem might have
to do with using swap space. I can restart it later, now that I know that
it is just running that input file, but I don't know whether, after
r21bugsbig.input runs (and presumably r21bugs.input, all the other
files *bugs*.input having corresponding output files), there is anything
else that make has to do. Where are we in the Makefile when r21bugsbig.input
is being fed to Axiom?
Question: if one wants to define something directly in Lisp and make it
available in top level, where in the documentation is it explained how to
do that? In my experience, this is the most poorly explained aspect of
every computer algebra system that depends on a Lisp substrate. Good
as it usually is, the user interface is actually a part of the problem.

Ignorantly,
Allan Adler <ara@zurich.csail.mit.edu>
* Disclaimer: I am a guest and *not* a member of the MIT CSAIL. My actions and
* comments do not reflect in any way on MIT. Also, I am nowhere near Boston. 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: a limit wrong in Maple...



A very good catch. Actually, all Maple versions
at least since 1994 have the same bug.
limit(f(n)*exp(n), n= infinity);
 (2004) Maple 9.5.2 
0
 (2004) Maple 9.5 
0
 (2003) Maple 9 
0
 (2002) Maple 8 
0
 (2001) Maple 7 
0
 (2000) Maple 6 
0
 (1997) Maple V Rel 5 
0
 (1995) Maple V Rel 4 
0
 (1994) Maple V Rel 3 
0

What about other computer algebra systems?
AXIOM 3.1.4
Quote:  f := operator 'f; limit(f(n)*exp(n), n= %plusInfinity)
"failed" 
Mathematica 5.1
Quote:  Limit[f[n] Exp[n], n> Infinity]
Limit[f[n] Exp[n], n > Infinity] 
Maxima 5.9.1
Quote:  LIMIT(f(n)*EXP(n), n, INF);
LIMIT(f(n)*EXP(n), n, INF); 
MuPAD 3.1
Quote:  limit(f(n)*exp(n), n= infinity);
limit(f(n)*exp(n), n = infinity) 
In contrast, we have the same bug here.
Derive 6.1
Quote:  f(n) :=
LIM(f(n)*EXP(n), n, inf)

0
XCAS 0.5.0
Quote:  limit(f(n)*exp(n), n= +infinity)
0 
YACAS 1.0.54
Quote:  Limit(n,Infinity)f(n)*Exp(n)
0 
Best wishes,
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/ 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: a limit wrong in Maple...



A very good catch. Actually, all Maple versions
at least since 1994 have the same bug.
limit(f(n)*exp(n), n= infinity);
 (2004) Maple 9.5.2 
0
 (2004) Maple 9.5 
0
 (2003) Maple 9 
0
 (2002) Maple 8 
0
 (2001) Maple 7 
0
 (2000) Maple 6 
0
 (1997) Maple V Rel 5 
0
 (1995) Maple V Rel 4 
0
 (1994) Maple V Rel 3 
0

What about other computer algebra systems?
AXIOM 3.1.4
Quote:  f := operator 'f; limit(f(n)*exp(n), n= %plusInfinity)
"failed" 
Mathematica 5.1
Quote:  Limit[f[n] Exp[n], n> Infinity]
Limit[f[n] Exp[n], n > Infinity] 
Maxima 5.9.1
Quote:  LIMIT(f(n)*EXP(n), n, INF);
LIMIT(f(n)*EXP(n), n, INF); 
MuPAD 3.1
Quote:  limit(f(n)*exp(n), n= infinity);
limit(f(n)*exp(n), n = infinity) 
In contrast, we have the same bug here.
Derive 6.1
Quote:  f(n) :=
LIM(f(n)*EXP(n), n, inf)

0
XCAS 0.5.0
Quote:  limit(f(n)*exp(n), n= +infinity)
0 
YACAS 1.0.54
Quote:  Limit(n,Infinity)f(n)*Exp(n)
0 
Best wishes,
Vladimir Bondarenko
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/ 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Maple 9 is unable to solve numerically this equation but Maple 5 does it!



hjbor...@mat.pucrio.br (Humberto Jose Bortolossi) writes on Mon,
Feb 14 2005 9:51 am
HJB> Maple 9 is unable to execute this:
HJB> fsolve(min(x+51580800.0,5652700000.0,x+51580800.0+.858535868
HJB> 2e13/x^(1/2))=4368842661., x, 1684100000.0 .. 5652700000.0);
HJB> On the other hand, Maple V and Maple 6 does it well.
Thank you for bringing to our attention your results.
The behavior you share with us the Maplesoft customers, is
TYPICAL.
According to our computational results, part of them coming in the
immediate future, there are at least thousands regression bugs in
Maple.
As for your fsolve example, only Maple V Release 3 of 1994 and
Maple V Release V of 1997 approximate the solution correctly.
Please especially note the coming and going INSTABILITY in Maple
9.5.2, Maple 9.5, and Maple 6, absent in Maple 9/8/7, and Maple V
Release 4.
To reproduce the data below, just copy the input some 2030 times
and see how Maple returns now 'Error, blahblahblah'
now '4611769259.'
restart; fsolve(min(x+51580800.0,5652700000.0,x+51580800.0+
..8585358682e13/x^(1/2))= 4368842661., x, 1684100000.0 ..
5652700000.0);
Maple 9.5.2
SIDE EFFECT: being asked repeatedly, Maple returns 2 distinct
answers for the same input.
4611769259.
Error, (in PiecewiseTools:Convert) unable to convert
Maple 9.5
SIDE EFFECT: being asked repeatedly, Maple returns 2 distinct
answers for the same input.
4611769259.
Error, (in PiecewiseTools:Convert) unable to convert
Maple 9
Error, (in PiecewiseTools:Convert) unable to convert
Maple 8
Error, (in PiecewiseTools:Convert) unable to convert
Maple 7
Error, (in PiecewiseTools:Convert) unable to convert
Maple 6
SIDE EFFECT: being asked repeatedly, Maple returns 2 distinct
answers for the same input.
4611769259.
Error, (in unitstep/red) cannot determine positive portion of unitstep
Maple V Rel 5
4611769259.
Maple V Rel 4
Error, (in type/realcons) too many levels of recursion
Maple V Rel 3
4611769259.

Best wishes,
Vladimir Bondarenko
GEMM architect
Cofounder, CEO, Mathematical Director
Cyber Tester, LLC
13 Dekabristov Str, Simferopol
Crimea 95000, Ukraine
tel: +38(0652)447325
tel: +38(0652)230243
tel: +38(0652)523144
fax: +38(0652)510700
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/ 

Back to top 


meznaric science forum beginner
Joined: 24 Mar 2005
Posts: 1

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: (",) Do You Want To Know For Sure You Are Going To Heaven?



Can you please not post such messages on this board? Hell, are these
people paid to do this? Moderators!? 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Regular search facility is absent at Maplesoft site during 3+ months, why?



Hi all,
Since October 2004, for more than 3 months, I keep trying to
use a regular search form at the main Maplesoft site...
No luck yet.
I keep seeing at http://www.maplesoft.com/search/index.aspx
the same message
"The Maplesoft sitewide search feature is currently being
redeveloped and will soon be available to help you."
I believe here we are running into 2 possibilities:
1. A grandiose masterpiece of Search Art is coming within
the next months.
2. Maplesoft has no idea of providing this regular facility
whatsoever, and the above message is no more than yet another
Maplesoft's blind, yet another unfulfilled promise, as it is
unfortunately became regular for Maplesoft during the last
years.
Also, I fully agree with Fred's opinion
http://groupsbeta.google.com/group/sci.math.symbolic/msg/023087adaae30f64
that full Maple documentation should be available online
(like for example, Wolfram Research, Inc, the Mathematica
makers
http://documents.wolfram.com/mathematica/
and SciFace GmbH, the MuPAD makers
http://research.mupad.de/doc/
have done it long ago) as well as with his comments about
the location/size of the Search button.
F> On maple main site, even the search button itself is hard
F> to see, small and hidden, and one must click on it first
F> to go to another page where you are supposed to be able
F> to write the search string (instead of typing the search
F> string right there on the main page as with the MMA case).
F> And when one goes to the search page, there is no place
F> to put a search string and one have to click on links
F> to try to find what THEY want.
Yes, what THEY want for OUR money.
Best wishes,
Vladimir Bondarenko
VM and GEMM architect
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/
Fred wrote:
Quote:  I found Maplesoft site not to be well designed at all.
Try to search for documentation for example about Maple
functions. There is none.
Someone who never used Maple and wants to get an idea of
its capabilities, then the first thing one would do is
look for a list of the functions provided. There is no
such list.
Compare that to the Mathematica web site, go to main site,
type documentation in the search box, first entry shown is
http://documents.wolfram.com/mathematica/
and there is a list of all the build in functions. with
FULL documentation of each function, and examples.
On maple main site, even the search button itself is hard to
see, small and hidden, and one must click on it first to go
to another page where you are supposed to be able to
write the search string (instead of typing the search
string right there on the main page as with the MMA case).
And when one goes to the search page, there is no place to
put a search string and one have to click on links to
try to find what they want.
Vladimir Bondarenko wrote:
Hello all,
The thrilling riddle of the Maplesoft Site Search heat
my imagination.
During the last month I almost every day visit Maplesoft and
at any attempt to use the Search facility at
http://www.maplesoft.com/search/
see the same:
The Maplesoft sitewide search feature is currently being
redeveloped and will soon be available to help you. In the
meantime, you can search through the Application Center and
Maple books databases.
No, I need not these, I need just a regular search facility.
What a wonder of software engineering Art the user will see
one fine day that required so much time to be designed and
implemented?
But most of all, a terrible suspicion gnaws my heart.
Will it also have a Java interface?! ;)
Oh noooo, God forbid!
Best wishes,
Vladimir Bondarenko
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/



Back to top 


Paul Abbott science forum addict
Joined: 19 May 2005
Posts: 99

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Bug in mathematica? or me?



In article <Y7GSd.88$Zn2.48@newsfe5gui.ntli.net>,
"Tony King" <mathstutoring@ntlworld.com> wrote:
Quote:  Please see attached notebook
I believe that the answer to this calculation is
68689590687165743191639263603153168999. However, either I am not fully
understanding the way that mathematica works internally or there is a
bug in it. Any comments would be helpful

The answer is _not_ 68689590687165743191639263603153168999. Here is the
correct answer:
(Sqrt[6]/24) (Sqrt[3] + Sqrt[2])^(4 n  2) /. n > 20 // ExpandAll
From the form of this result you can see that the answer is _not_ an
integer but involves Sqrt[6].
This answer differs from 6868959068716574319163926360315316899:
N[%  68689590687165743191639263603153168999, 100]
The difference is small (~10^(40)) but definitely not zero.
Cheers,
Paul

Paul Abbott Phone: +61 8 6488 2734
School of Physics, M013 Fax: +61 8 6488 1014
The University of Western Australia (CRICOS Provider No 00126G)
35 Stirling Highway
Crawley WA 6009 mailto:paul@physics.uwa.edu.au
AUSTRALIA http://physics.uwa.edu.au/~paul 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Turn 6 bucks into a small fortune fast with PayPal



Thank you kindly for sharing with us your revolutionary, very
very dependable  and fully legal, this is best!  way of
making of long greens.
It would be great if you inform Maplesoft about your outstanding
idea of making high ROI.
Personally, I have a bad feeling that had Maplesoft really
choose further ignoring serious work on quality, the way you
propose could become the only way of making money for this
company.
Because, in particular, the competitors are not in the winter
hibernation at all. For example, I'd like to state my clear
satisfaction with the drive Wolfram Research, Inc started
lately fixing their bugs in various functions.
For more details read the Maple Review coming.
Best wishes,
Vladimir Bondarenko
VM and GEMM architect
Cofounder, CEO, Mathematical Director
Cyber Tester, LLC
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/ 

Back to top 


Gnu_Raiz science forum beginner
Joined: 24 Mar 2005
Posts: 1

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Regular search facility is absent at Maplesoft site during 3+ months, why?



On Tue, 01 Mar 2005 23:48:23 0800, Vladimir Bondarenko wrote:
Quote:  Hi all,
Since October 2004, for more than 3 months, I keep trying to
use a regular search form at the main Maplesoft site...
No luck yet.
I keep seeing at http://www.maplesoft.com/search/index.aspx
the same message
"The Maplesoft sitewide search feature is currently being
redeveloped and will soon be available to help you."
I believe here we are running into 2 possibilities:
1. A grandiose masterpiece of Search Art is coming within
the next months.
2. Maplesoft has no idea of providing this regular facility
whatsoever, and the above message is no more than yet another
Maplesoft's blind, yet another unfulfilled promise, as it is
unfortunately became regular for Maplesoft during the last
years.
snip
have done it long ago) as well as with his comments about
the location/size of the Search button.
F> On maple main site, even the search button itself is hard
F> to see, small and hidden, and one must click on it first
F> to go to another page where you are supposed to be able
F> to write the search string (instead of typing the search
F> string right there on the main page as with the MMA case).
F> And when one goes to the search page, there is no place
F> to put a search string and one have to click on links
F> to try to find what THEY want.
Yes, what THEY want for OUR money.
Best wishes,
Vladimir Bondarenko
snip 
I don't see what the problem is, other than your habit of top posting,
and html in you signature file. I hope you will be more considerate in the
future and follow proper posting procedures.
http://www.netmeister.org/news/learn2quote2.html
To solve your search problem you might want to consider using google's
search site feature, like this as an example. This is all done in google's
search bar, just type this in, or what ever your searching for.
truth tables site:www.maplesoft.com
This will return information only on maplesofts site, this is handy on
other sites that have large volumes of information, or poor search
engines.
Gnu_Raiz 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Regular search facility is absent at Maplesoft site during 3+ months, why?



Gnu_Raiz <Gnu_R...@uptime.notlost.net> wrote on Sat, Mar 12 2005 9:49
am
GR> I don't see what the problem is,
Let me explain specially for you:
With the current inadequate support of Maple and Maple customers,
with its current ostrich policy of noncommunicating with its
customers (in contrast with, for example, Wolfram Research and
SciFace GmbH customersdevelopers dialogues) Maplesoft is running
real risk of going out of business, and Cyber Tester would like to
attract the attention of the Maplesoft's leaders and general public
to this sad possible outcome.
GR> other than your habit of top posting,
This has nothing to do with habits, mine or someone's.
This is a question of strategy.
GR> http://www.netmeister.org/news/learn2quote2.html
Thank you for the link.
GR> To solve your search problem
I have NO problem with search whatsoever.
It is Maplesoft which has this problem... and not only this.
GR> you might want to consider using google's search site feature
If you really think I couldn't find a workaround, you may wish to
visit
http://www.castesting.org/index.php?list=3
http://www.castesting.org/index.php?list=7
http://www.castesting.org/SciFace.phtml
etc
The problem is that the behavior Maplesoft demonstrates with the
current longsuspended status of their Search is unfortunately
typical for the most recent period of Maple development, and the
exact name for it is,
LACK OF ERGONOMICS.
In other words, the Maple customers buy Maple to resolve our tasks
AS QUICKLY AS IT IS POSSIBLE. But alas these hopes fail too often,
and something should be done to fix this plight.
http://groupsbeta.google.com/group/sci.math.symbolic/msg/dd5e6a02c428dad9
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/6002b726f9f2e669
http://groupsbeta.google.com/group/comp.softsys.math.maple/search?group=comp.softsys.math.maple&q=vb@cybertester.com&qt_g=1&searchnow=Search+this+group
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/5067451c29b50714
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/691d8f9b4295cb82
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/de2098727d9fed4f
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/860d5190bcff0bba
http://groupsbeta.google.com/group/sci.math.symbolic/msg/9e2bd397c9c9e9fd
etc etc
Best wishes,
Vladimir Bondarenko
VM and GEMM architect
Cofounder, CEO, Mathematical Director
Cyber Tester, LLC
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/ 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: Regular search facility is absent at Maplesoft site during 3+ months, why?



Could you please repost 'MuPAD help menu?' as a separate
thread so it would be more visible, and does not shade an
unrelated thread?
Thank you in advance. 

Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: euler(0,1) shows Maple bugs are really ubiquitous



Hi Maple customers all over the world,
Happy birthday,
happy birthday,
happy birthday to you,
o stupid Maple bug!
Many happy returns of the day!
Laurent Bernardin, a Chief Scientist and VP, Research and
Development, Adjunct Professor, Dept. of Pure Mathematics,
University of Waterloo. claims:
http://bernardin.com/maple/index.php
LB> Being strict about backwards compatibility can be frustrating
LB> for our developers. A common comment I hear, is: "It would
LB> really be easier to make this command work better, if I would
LB> not have to worry about compatibility". This is a good sign.
LB> It means the issue is taken seriously at all levels. It also
LB> means that we are finding ways to move Maple forward and
LB> improve the system without breaking existing user code.
Peter Luschny proposed on Dec 23 2003 a splendid idea
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/572265481f09ebe5
PL> Vladimir's marketing is not perfect, yet. But what
PL> about the slogan: "Maple  unsafe at any evaluation"?
Oh yes!
This is a hot piece of news on 2003 Peter's remarkable
extended comments.
PL> If we have to start to fix even the constant functions
PL> in Maple first, we can very well refrain from using
PL> Maple at all.
PL> Second: Can you tell us how many software out there,
PL> build since
PL> Maple V, Release 3, IBM INTEL NT, Jan 10, 1994
PL> ** 10 years now, happy birthday **
PL> use explicit or implicit this function and is
PL> by this very fact buggy? How can you help them?
PL> You can't.
PL> Most of them will never be aware of this fact but
PL> all of them might be bitten some day by the bug.
PL> And third: I did not really expect a bug fix, but a
PL> comment, what might be so fundamentally wrong at
PL> Maplesoft, that they cannot even guarantee a
PL> constant function to have a constant value.
PL> The usual blahblah of inherent complexity etc.
PL> used to exculpate inability, is obviously not
PL> applicable here.
After 10 years of its existent, in Maple 9.5.2 the bug with
euler(0,1) reported by Peter was fixed...
euler(0,1);
 (2004) Maple 9.5.2 
1
 (2004) Maple 9.5 
1
 (2003) Maple 9 
1
 (2002) Maple 8 
1
 (2001) Maple 7 
1
 (2000) Maple 6 
1
 (1997) Maple V Rel 5 
1
 (1995) Maple V Rel 4 
1
 (1994) Maple V Rel 3 
1

Was fixed... partially!
And this (partial, juryrigged, quickfix) approach is TYPICAL for
Maplesoft. Read the Maple Review coming very soon for details.
Here is the bug, alive and kickin'!
limit(euler(z,1), z=0);
 (2004) Maple 9.5.2 
1
 (2004) Maple 9.5 
1
 (2003) Maple 9 
1
 (2002) Maple 8 
1
 (2001) Maple 7 
1
 (2000) Maple 6 
1
 (1997) Maple V Rel 5 
1
 (1995) Maple V Rel 4 
1
 (1994) Maple V Rel 3 
1

http://www.maplesoft.com/
Maplesoft> Maple 9.5: Standard for any type of mathematics
Hehe... A zero cool slogan. The only rub is that it is
not implemented in real life.
Read much more about the 'standard' Maplesoft proposes us
to use, in the immediate future.
All this is just the very beginning,
Vladimir Bondarenko
VM and GEMM architect
Cofounder, CEO, Mathematical Director
Cyber Tester, LLC
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/
Peter Luschny wrote:
Quote:  "Alec Mihailovs" wrote:
"Vladimir Bondarenko"
PL> euler(0,1); # Sure, this must be equal to 1
1
As usual, this can be easily fixed:
myeuler:=n> if n=0 then 1 else euler(args) fi:
myeuler(0,1);
I am impressed Thanks.
But, as usual, you did not get the point.
If we have to start to fix even the constant functions
in Maple first, we can very well refrain from using
Maple at all.
Second: Can you tell us how many software out there,
build since
Maple V, Release 3, IBM INTEL NT, Jan 10, 1994
** 10 years now, happy birthday **
use explicit or implicit this function and is
by this very fact buggy? How can you help them?
You can't.
Most of them will never be aware of this fact but
all of them might be bitten some day by the bug.
And third: I did not really expect a bug fix, but a
comment, what might be so fundamentally wrong at
Maplesoft, that they cannot even guarantee a
constant function to have a constant value.
The usual blahblah of inherent complexity ect.
used to exculpate inability, is obviously not
applicable here.
To build a robust, all time correct code for
euler(0,x) all you have to do is
euler := proc(n,x) if n=0 then RETURN(1) else..
(Which is even shorter than your bug fix.)
They obviously use other techniques, and these
techniques of softwaredevelopment are not robust
and not state of the art, use techniques which fail
over and over again.
They are to be asked why.
What I can imagine as a possible cause: The
developers there know too much of mathematics
and not enough of software development.
And the management fails to realize this.
The Javafrontend debacle might indicate that
the management cannot even identify competent
developers for a given task.
Regards Peter 


Back to top 


Vladimir Bondarenko science forum Guru
Joined: 23 Apr 2005
Posts: 601

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: euler(0,1) shows Maple bugs are really ubiquitous



Hi Maple customers all over the world,
Happy birthday,
happy birthday,
happy birthday to you,
o stupid Maple bug!
Many returns of the day!
Laurent Bernardin, a Chief Scientist and VP, Research and
Development, Adjunct Professor, Dept. of Pure Mathematics,
University of Waterloo. claims:
http://bernardin.com/maple/index.php
LB> Being strict about backwards compatibility can be frustrating
LB> for our developers. A common comment I hear, is: "It would
LB> really be easier to make this command work better, if I would
LB> not have to worry about compatibility". This is a good sign.
LB> It means the issue is taken seriously at all levels. It also
LB> means that we are finding ways to move Maple forward and
LB> improve the system without breaking existing user code.
Peter Luschny proposed on Dec 23 2003 a splendid idea
http://groupsbeta.google.com/group/comp.softsys.math.maple/msg/572265481f09ebe5
PL> Vladimir's marketing is not perfect, yet. But what
PL> about the slogan: "Maple  unsafe at any evaluation"?
Oh yes!
This is a hot piece of news on 2003 Peter's remarkable
extended comments.
PL> If we have to start to fix even the constant functions
PL> in Maple first, we can very well refrain from using
PL> Maple at all.
PL> Second: Can you tell us how many software out there,
PL> build since
PL> Maple V, Release 3, IBM INTEL NT, Jan 10, 1994
PL> ** 10 years now, happy birthday **
PL> use explicit or implicit this function and is
PL> by this very fact buggy? How can you help them?
PL> You can't.
PL> Most of them will never be aware of this fact but
PL> all of them might be bitten some day by the bug.
PL> And third: I did not really expect a bug fix, but a
PL> comment, what might be so fundamentally wrong at
PL> Maplesoft, that they cannot even guarantee a
PL> constant function to have a constant value.
PL> The usual blahblah of inherent complexity etc.
PL> used to exculpate inability, is obviously not
PL> applicable here.
After only 10 years of its existence, in Maple 9.5.2 the bug
with euler(0,1) reported by Peter was fixed...
euler(0,1);
 (2004) Maple 9.5.2 
1
 (2004) Maple 9.5 
1
 (2003) Maple 9 
1
 (2002) Maple 8 
1
 (2001) Maple 7 
1
 (2000) Maple 6 
1
 (1997) Maple V Rel 5 
1
 (1995) Maple V Rel 4 
1
 (1994) Maple V Rel 3 
1

Was fixed... partially!
And this (partial, juryrigged, quickfix) approach is TYPICAL for
Maplesoft. Read the Maple Review coming very soon for details.
Here is the bug, alive and kickin'!
limit(euler(z,1), z=0);
 (2004) Maple 9.5.2 
1
 (2004) Maple 9.5 
1
 (2003) Maple 9 
1
 (2002) Maple 8 
1
 (2001) Maple 7 
1
 (2000) Maple 6 
1
 (1997) Maple V Rel 5 
1
 (1995) Maple V Rel 4 
1
 (1994) Maple V Rel 3 
1

http://www.maplesoft.com/
Maplesoft> Maple 9.5: Standard for any type of mathematics
Hehe... A zero cool slogan. The only rub is that it is
not implemented in real life.
Read much more about the 'standard' Maplesoft proposes us
to use, in the immediate future.
All this is just the very beginning,
Vladimir Bondarenko
VM and GEMM architect
Cofounder, CEO, Mathematical Director
Cyber Tester, LLC
http://www.cybertester.com/
http://maple.buglist.org/
http://www.CAStesting.org/
Peter Luschny wrote:
Quote:  "Alec Mihailovs" wrote:
"Vladimir Bondarenko"
PL> euler(0,1); # Sure, this must be equal to 1
1
As usual, this can be easily fixed:
myeuler:=n> if n=0 then 1 else euler(args) fi:
myeuler(0,1);
I am impressed Thanks.
But, as usual, you did not get the point.
If we have to start to fix even the constant functions
in Maple first, we can very well refrain from using
Maple at all.
Second: Can you tell us how many software out there,
build since
Maple V, Release 3, IBM INTEL NT, Jan 10, 1994
** 10 years now, happy birthday **
use explicit or implicit this function and is
by this very fact buggy? How can you help them?
You can't.
Most of them will never be aware of this fact but
all of them might be bitten some day by the bug.
And third: I did not really expect a bug fix, but a
comment, what might be so fundamentally wrong at
Maplesoft, that they cannot even guarantee a
constant function to have a constant value.
The usual blahblah of inherent complexity ect.
used to exculpate inability, is obviously not
applicable here.
To build a robust, all time correct code for
euler(0,x) all you have to do is
euler := proc(n,x) if n=0 then RETURN(1) else..
(Which is even shorter than your bug fix.)
They obviously use other techniques, and these
techniques of softwaredevelopment are not robust
and not state of the art, use techniques which fail
over and over again.
They are to be asked why.
What I can imagine as a possible cause: The
developers there know too much of mathematics
and not enough of software development.
And the management fails to realize this.
The Javafrontend debacle might indicate that
the management cannot even identify competent
developers for a given task.
Regards Peter 


Back to top 


Peter Luschny science forum beginner
Joined: 13 May 2005
Posts: 36

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: euler(0,1) shows Maple bugs are really ubiquitous



"Vladimir Bondarenko" tested yesterday:
Quote:  limit(euler(z,1), z=0);

Oh, what an interesting question are you investigating!
Are these the confluent Euler Polynomials?
Well, I do not know how Maplesoft defines euler(s,z),
s not an integer, therefore I will take my own approach.
Quote:  restart;
E := proc(s,z) z^s*(4^(s+1)*Zeta(0,s,1/(4*z))
 2^(s+1)*Zeta(0,s,1/(2*z))) end:
`E(s,z)`=E(s,z);

Let's get a first impression what this looks like for s=0:
Quote:  plot(E(s,1),s=1/2..1/2);

Seems as if E(0,1) = 1. And indeed:
Fine! So let us compute next:
Quote:  limit(E(s,1),s=0,real);
2 
Uhh, now I am confused. :/
Well, I am using Maple V Release 5,
perhaps I should upgrade?
Regards Peter 

Back to top 


A N Niel science forum Guru
Joined: 28 Apr 2005
Posts: 475

Posted: Thu Mar 24, 2005 7:46 pm Post subject:
Re: euler(0,1) shows Maple bugs are really ubiquitous



In article <39o0iqF648iejU1@individual.net>, Peter Luschny
<peter.luschny@gmx.net> wrote:
Quote:  "Vladimir Bondarenko" tested yesterday:
limit(euler(z,1), z=0);
Oh, what an interesting question are you investigating!
Are these the confluent Euler Polynomials?
Well, I do not know how Maplesoft defines euler(s,z),
s not an integer, therefore I will take my own approach.

According to ?euler ...
euler(n, x)
n  nonnegative integer
x  expression
So it seems putting in a noninteger is not supported...
Quote:  euler(1/4,x);
Error, (in euler) invalid arguments 
Quote: 
restart;
E := proc(s,z) z^s*(4^(s+1)*Zeta(0,s,1/(4*z))
 2^(s+1)*Zeta(0,s,1/(2*z))) end:
`E(s,z)`=E(s,z);
Let's get a first impression what this looks like for s=0:
plot(E(s,1),s=1/2..1/2);
Seems as if E(0,1) = 1. And indeed:
E(0,1);
1
Fine! So let us compute next:
limit(E(s,1),s=0,real);
2
Uhh, now I am confused. :/

Here's _your_ problem:
Quote:  restart;Zeta(0,1,1/4);limit(Zeta(0,x,1/4),x=1);

1

96
1

12
Maple 9.52.
The mitigation I find is in ?Zeta...
Zeta(n, z, v)
v  algebraic expression, understood not to be a nonpositive
integer
and 1/4 isn't an integer. But:

12
so we were given the value at a nearby integer ... ???
I guess that's our punishment for going outside the intended domain of
definition. 

Back to top 


Google


Back to top 



The time now is Mon Dec 10, 2018 10:24 pm  All times are GMT

