vladimir bondarenko

.................................................. ..............

A tiny demo example of efficiency of the VM machine.

.................................................. ..............

In[1] := $Version

Out[1] = 5.0 for Microsoft Windows (June 11, 2003)

In[2] := Limit[z, z -> Exp[Exp[-Exp[10]]]-1]

Out[2] = 0


The answer should be, of course, Exp[Exp[-Exp[10]]]-1 .

.................................................. ..............

Subj: The MathKernel can be reinforced (Cover Letter)

http://groups.google.com/group/sci.math.symbolic/browse_frm/thread/a4ca5a4a42c323a5/#

.................................................. ..............

Vladimir Bondarenko

VM and GEMM architect
Co-founder, CEO, Mathematical Director

http://www.cybertester.com/ Cyber Tester, LLC
http://maple.bug-list.org/ Maple Bugs Encyclopaedia
http://www.CAS-testing.org/ CAS Testing

.................................................. ..............

peter pein

Vladimir Bondarenko schrieb:

Mathematica should give an error message or warning about insufficient
precision in situations like this (and usually does).


$MaxExtraPrecision = 9543;
Limit[z, z -> Exp[Exp[-Exp[10]]] - 1]
--> -1 + E^E^(-E^10)

Peter

vladimir bondarenko

Peter Pein wrtes:

PP> Mathematica should give an error message or warning
PP> about insufficient precision in situations like this

Yes, I am absolutely agree with you. Wolfram Research Inc's
customers enjoy a unique streamlined powerful machinery;
usually it tracks down the insufficient precision problems
which the warnings serve a good advising and helping role
allowing the customers to recognize potential incorrect
answers.

PP> (and usually does).

Yes, this is a nice mechanism and usually it works very
well. The more credit should be given to the VM machine
which still can trap successfully such problems.

A moment I like much in Wolfram Research is that they are
very active about the reported defects in Mathematica.

Their enviable programming and research experience and
alertness about the quality multiplied by solid proactive
attitude is a good weapon against the defects.

For example, we can enjoy an amazing activity a key WRI
person, Dr Daniel Lichtblau displays in Google Groups.


Cheers,

Vladimir

rjf

The only thing eerie about this is that the computation apparently
relies on some heuristic, when it shouldn't. It isn't a sufficient
mechanism for this problem and therefore shouldn't be used. How VB
comes to the conclusion that somehow this is evidence of Mathematica's
level of support is beyond me.

I think that DanL's contributions to the community are worthly of high
praise, but that does not have much bearing on whether this is a bug or
not.