Re: Integrating tan(x)/x
- From: stevendaryl3016@xxxxxxxxx (Daryl McCullough)
- Date: 23 Apr 2007 07:54:54 -0700
In article <1177076761.537782.314520@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>, *** says...
Hi all,
Does anybody know how the integral from 0 to infinity of tan(x)/x
get assigned the value of pi/2? I can't think of any way to naturally
assign a contour or of any way to manipulate Si(x) to yield it. Just
curious.
Thanks, ***
G. A. Edgar already gave the contour integral way of solving
the problem, but just for variety, here's the way that I did
it:
Note Integral from x=0 to infinity of tan(x)/x can
be written as
sum from n=0 to infinity of
Integral from x=0 to pi/2 tan(x + n pi)/(x+n pi) dx
+ Integral from x=pi/2 to pi of tan(x + n pi)/(x + n pi) dx
If we use tan(x+n pi) = tan(x) and
tan(pi - x) = - tan(x), we can rewrite this as
sum from n=0 to infinity of
Integral from x=0 to pi/2 tan(x)/(x+n pi) dx
+ Integral from x=0 to pi/2 of -tan(x)/((n+1) pi - x) dx
where in the second integral, I made the substitution
x --> pi - x. Assuming that it is permissable to interchange
summation with integration (which requires a proof, but I'm
not going to give it), this can be rewritten as
Integral from x=0 to pi/2 of tan(x) * f(x) dx
where f(x) = 1/x + 1/(pi + x) - 1/(pi - x) + 1/(2 pi + x) - 1/(2 pi - x) ...
= sum from n=-infinity to +infinity of 1/(x - n pi)
(where the infinite sum is the limit of the sum from n=-K to +K
as K --> infinity).
Since it is an alternating series, it will converge, provided that
x is not equal to a multiple of pi. What does it converge to?
Amazingly enough, it converges to cot(x). That's plausible, since
it has poles at all the right places (x = 0, +/- pi, +/- 2 pi, etc.).
To prove that f(x) = cot(x), one way is to start with the product
representation of sin(x), written in a special way:
sin(x) = x * 1/pi * (pi + x)
* 1/pi * (pi - x)
* 1/(2 pi) * (2 pi + x)
* 1/(2 pi) * (2 pi - x)
* ...
Then log(sin(x)) = log(x)
+ log(1/pi) + log(pi + x)
+ log(1/pi) + log(pi - x)
+ log(1/(2 pi)) + log(2pi + x)
+ log(1/(2 pi)) + log(2pi - x)
+ ...
Then d/dx log(sin(x))
= 1/x + 1/(pi + x) - 1/(pi - x) + 1/(2pi + x) - 1/(2pi - x) + ...
So cot(x) = 1/x + 1/(pi + x) - 1/(pi - x) + 1/(2pi + x) - 1/(2pi - x) + ...
--
Daryl McCullough
Ithaca, NY
.
- References:
- Integrating tan(x)/x
- From: ***
- Integrating tan(x)/x
- Prev by Date: Re: Fermat's Last theorem short proof
- Next by Date: Re: What Is a Mathematician?
- Previous by thread: Re: Integrating tan(x)/x
- Next by thread: Transform method for PDEs
- Index(es):