Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.soft-sys.math.mathematica > #13129 > unrolled thread

Re: wrong result when computing a definite integral

Started byAlex Krasnov <akrasnov@eecs.berkeley.edu>
First post2013-01-14 05:01 +0000
Last post2013-01-14 05:01 +0000
Articles 1 — 1 participant

Back to article view | Back to comp.soft-sys.math.mathematica

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: wrong result when computing a definite integral Alex Krasnov <akrasnov@eecs.berkeley.edu> - 2013-01-14 05:01 +0000

#13129 — Re: wrong result when computing a definite integral

FromAlex Krasnov <akrasnov@eecs.berkeley.edu>
Date2013-01-14 05:01 +0000
SubjectRe: wrong result when computing a definite integral
Message-ID<kd03eu$6du$1@smc.vnet.net>
I should have probably read the problem before posting. There should be no 
issue with branches, since Exp is single-valued. Interestingly, the 
incorrect result differs by a phase factor of (-2+Sqrt[3])*Pi.

I also noticed that the documentation states that Integrate computes 
multiple integrals. It actually computes interated integrals in prefix 
notation:

Integrate[f, y, x] <=> Integrate[dy*Integrate[dx*f]]

This is clear from the following example:

In:	Integrate[(x^2-y^2)/(x^2+y^2)^2, {y, 0, 1}, {x, 0, 1}]
Out:	-Pi/4

In:	Integrate[(x^2-y^2)/(x^2+y^2)^2, {x, 0, 1}, {y, 0, 1}]
Out:	Pi/4

Since multiple and iterated integrals are equal only through Fubini's 
theorem and similar results, perhaps the documentation should be 
corrected.

Alex


On Sat, 12 Jan 2013, Murray Eisenberg wrote:

> Unless there's some issue of branches of complex functions involved that
> I'm missing, it should not matter here which order of integration you
> use -- since the limits of integration are constants. However, if you
> wrap each integrand in ComplexExpand,
>
>   a = Integrate[ComplexExpand[Exp[I*Sqrt[3]*y]], {x, -2*Pi, 2*Pi},
> {y, -Pi, Pi}]
>   b = Integrate[ComplexExpand[Exp[I*Sqrt[3]*y]], {y, -Pi, Pi}, {x,
> -2*Pi, 2*Pi}]
>
> then you obtain the same result:
>
>   {a, b} // InputForm
> {(8*Pi*Sin[Sqrt[3]*Pi])/Sqrt[3], (8*Pi*Sin[Sqrt[3]*Pi])/Sqrt[3]}
>   a == b
> True
>
> On Jan 11, 2013, at 10:22 PM, Alex Krasnov <akrasnov@eecs.berkeley.edu> wrote:
>
>> Integrate takes the integration variables in prefix order, so perhaps you
>> meant the following:
>>
>> In:	Integrate[Exp[I*Sqrt[3]*y], {y, -Pi, Pi}, {x, -2*Pi, 2*Pi}]
>> Out:	(8*Pi*Sin[Sqrt[3]*Pi])/Sqrt[3]
>>
>>  Thu, 10 Jan 2013, Dexter Filmore wrote:
>>
>>> i run into this problem today when giving a bunch of easy integrals to mathematica.
>>> here's a wolfram alpha link to the problem:
>>>
> http://www.wolframalpha.com/input/?i=Integrate%5BExp%5BI+Sqrt%5B3%5Dy%5D%2C%7Bx%2C-2Pi%2C2Pi%7D%2C%7By%2C-Pi%2CPi%7D%5D#
>>>
>>> the integrand does not depend on the 'x' variable, the inner
> integration should only result in a factor of 4Pi, and the correct
> result is a real number, yet the below integral gives a complex number
> which is far off from the correct value:
>>> Integrate[Exp[I Sqrt[3] y], {x, -2 Pi, 2 Pi}, {y, -Pi, Pi}] -> -((4 I (-1 + E^(2 I Sqrt[3] Pi)) Pi)/Sqrt[3])
>>>
>>> from some trial and error it seems the result is also incorrect for non-integer factors in the exponential.
>
> ---
> Murray Eisenberg                                    murray@math.umass.edu
> Mathematics & Statistics Dept.
> Lederle Graduate Research Tower            phone 413 549-1020 (H)
> University of Massachusetts                               413 5 (W)
> 710 North Pleasant Street                         fax   413 545-1801
> Amherst, MA 01003-9305
>
>
>
>
>
>

[toc] | [standalone]


Back to top | Article view | comp.soft-sys.math.mathematica


csiph-web