Mathematica 9 is now available
Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
2003
*January
*February
*March
*April
*May
*June
*July
*August
*September
*October
*November
*December
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 2003

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

Search the Archive

Re: Re: Improper integral

  • To: mathgroup at smc.vnet.net
  • Subject: [mg44689] Re: [mg44667] Re: Improper integral
  • From: Andrzej Kozlowski <akoz at mimuw.edu.pl>
  • Date: Fri, 21 Nov 2003 05:13:12 -0500 (EST)
  • References: <6C03C616-19CE-11D8-864F-00039311C1CC@mimuw.edu.pl> <bpfftf$lvs$1@smc.vnet.net> <200311200816.DAA01569@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

On 20 Nov 2003, at 17:16, Jean-Claude Poujade wrote:

> Andrzej Kozlowski <akoz at mimuw.edu.pl> wrote in message 
> news:<bpfftf$lvs$1 at smc.vnet.net>...
>> More on this theme: can somebody explain what the concept of a
>> "principal value" of an integral is good for? I have been a
>> professional mathematician for years and have been involved in several
>> different areas of research, and yet never came across any use for it.
>> I have more then a dozen texts on analysis yet none of them mentions
>> it. The only books where I can find it mentioned are books for
>> physicists and engineers (one is the well known text by Riley, Hobson
>> and Bence, the other a book in Polish) and they both give one line
>> definitions without any examples of use (and do not mention poles at
>> infinity). At first sight it seems a pretty trivial and useless
>> concept, so I would like to know if it really has any serious
>> applications.
>>
>> Andrzej Kozlowski
>
> Andrzej,
>
> I'm sorry my subject was perhaps not enough explicit.
> I'm not arguing about the usefulness of the "principal value"
> concept : I just want to be allowed to remove as easily as
> possible some exceptions (for instance in probabilities)
> and I want Mathematica to comply with generally agreed definitions.
>
> Let me quote Borowski & Borwein's dictionary of mathematics :
>
> < Cauchy principal value : the evaluation of an improper integral
> < on the interval [-Infinity,Infinity] as the symmetric (two-sided)
> < limit of the integral on intervals of the form [-n,n].
> < This may well converge even if the sum of the two ordinary improper
> < integrals over [-Infinity,a] and [a,Infinity] does not.
>
> Given that definition, it seems to me that Mathematica doesn't return
> what it should. That's all I mean.
> ---
> jcp
>
>

In fact my question about the usefulness of concept of PrincipalValue 
of an integral was motivated by curiosity (stimulated by your original 
posting). I was surprised that I had (as it seemed to me) never come 
across any use of it. But that was not actually true; after Daniel 
Lichtblau pointed out that it was used in distribution theory I looked 
up one of the books that I once had read (more precisely skimmed 
through) and it was there, in some detail in fact. The principal value 
of an integral is actually itself an integral but not of a function but 
of a distribution.

The other question is whether Mathematica should have a full 
implementation of this. At this time it seems that it only computes 
principal values of integrals with singularities at points in the 
complex plane. In fact distribution theory uses precisely the principal 
values of integrals of the type that appeared in your example, that is, 
with singularities at infinity. The omission seems not to be an 
accident. I have looked at other cases at it appears that 
PrincipalValue for integrals from -Infinity to Infinity is not 
implemented in any case. Considering this more carefully one notices 
that the most obvious implementations of both kinds of  principal 
values ("finite" and "infinite") would conflict with one another.  I 
don't think it is an insurmountable difficulty but it seems to me that 
the current state of affairs is not a bug but a deliberate decision. 
One reason may be is that the principal value is so easy to compute 
(just take a suitable limit), that it does not appear worth the effort 
given that the problem of implementing both types of principal values 
is not entirely trivial.

Andrzej Kozlowski 


  • Prev by Date: Arg[JacobiSN + DrawContourLines
  • Next by Date: Conversion Function -> List
  • Previous by thread: Re: Improper integral
  • Next by thread: Re: Improper integral