What is the value of the following integral? $$\int_{0}^{\infty} \dfrac{1-\cos x}{x^{2}} dx$$
-
2This is a standard exercise in contour integration. Where did you encounter this, what have you tried, and where are you stuck? – Jul 17 '16 at 23:50
-
$\frac{\pi}{2}$ – Count Iblis Jul 17 '16 at 23:58
-
1"Evaluate" is an appropriate word; "solve" is not. One solves problems and one solves equations; one evaluates expressions. – Michael Hardy Jul 18 '16 at 02:31
-
I have a different solution using this answer. Please add some context so that your question can be answered. – robjohn Jul 18 '16 at 18:09
4 Answers
If you are aware of the rather well-known identity: $$ \forall \alpha>0,\qquad I(\alpha)=\lim_{N\to +\infty}\int_{0}^{N}\frac{\sin(\alpha x)}{x}\,dx = \frac{\pi}{2} \tag{1}$$ it is straightforward to check that: $$ J=\int_{0}^{+\infty}\frac{1-\cos(x)}{x^2}\,dx = \int_{0}^{1}I(\alpha)\,d\alpha = \color{red}{\frac{\pi}{2}}.\tag{2} $$ You may also use integration by parts to check that $J=I(1)$.

- 353,855
With the usual contour $\;C_{\epsilon,R}\;$ with a little "bump" around zero (canonical semicircle of radius $\;\epsilon\;$ and around the upper canonical semicircle of radius $\;R,\,\, R>>\epsilon\;$, and with the function
$$f(z)=\frac{1-e^{iz}}{z^2}\implies \text{Res}_{z=0}\,(f)= \lim_{z\to0} z\,f(z)=-ie^{i\cdot0}=-i$$implies
$$0=\lim_{\epsilon\to0,\,R\to\infty}\oint_{C_{\epsilon,R}}\,f(z)dz=\int_{-\infty}^\infty f(x)\,dx-\pi i(-i)\implies\int_0^\infty\frac{1-\cos x}{x^2}=\frac\pi2$$
by comparing real parts and noticing our function is even (and thus dividing by two)

- 211,718
- 17
- 136
- 287
Solution for lazy people. A special case of Ramanujan's master theorem was obtained by Glaisher in the late 19th century using non-rigorous methods. It says that if
$$f(x) = \sum_{k=0}^\infty (-1)^ka_k x^{2k}$$
then:
$$\int_0^\infty f(x) \, dx = \frac{\pi}{2} a_{-\frac{1}{2}}$$
if the integral here converges. Here one assumes a natural extrapolation of the series expansion coefficients that in the rigorous proof is made unambiguous. When this involves factorials, the prescription is to use the analytic continuation given by gamma functions. In this case, we have:
$$\frac{1-\cos(x)}{x^2} = \sum_{k=0}^\infty (-1)^k \frac{x^{2k}}{(2k+2)!}$$
therefore $a_{-\frac{1}{2}} = 1$ and the integral is thus equal to $\frac{\pi}{2}$.

- 10,366
You could get the answer via integration by parts for:
$$\int_0^\infty\frac{\cos(x)}{x^2} \, dx$$
via the below for suitably chosen $u(x)$ and $v(x)$;
$$\int_0^\infty v \, du = uv - \int_0^\infty u \, dv$$
Hence; $$\int_0^\infty \frac{\cos(x)}{x^2} \,dx= -\left[\cos(x) \left(\frac{-1}{x} \right) \right]_0^\infty - \int_0^\infty \frac{\sin(x)}{x}dx$$
Following Michael Hardy's comment;
$$\int_0^\infty \frac{-\sin(x)}{x} \, dx= -\left[\cos(x) \left( \frac{-1}{x} \right) \right]_0^\infty - \int_0^\infty \frac{-\cos(x)}{x^2}dx$$
We can now substitute this into the previous equation to find $\int_0^\infty \frac{\cos(x)}{x^2} \, dx$ in terms of $\cos(x)$ and $x$
For $\int_0^\infty \frac{1}{x^2} \, dx$, this is $\left[\dfrac{-1}{x}\right]_0^\infty$.

- 778
-
1I don't think integration by parts will help, neither in the integral you wrote and which is not the one the OP wrote, nor the actual one...but if I am wrong I'd appreciate you describe a little more what to do. – DonAntonio Jul 18 '16 at 00:10
-
I downvoted this because it's not just an unsupported assertion but quite questionable, to say the least. $$ \int \Big(\cos x\Big) \left( \frac{dx}{x^2}\right) = \int u,dv = uv - \int v, du $$ $$ = \frac{-\cos x} x - \int\frac{-1} x(-\sin x , dx) $$ Then what do you do after that? $\qquad$ – Michael Hardy Jul 18 '16 at 02:43
-
@DonAntonio - the difference of the two integrals in my answer is the integral the OP wrote. – unseen_rider Jul 18 '16 at 09:45
-
@michaelhardy then you can do the 2nd integral by parts again, giving you a relation for $\int_0^\infty \frac{cos(x)}{x^2}dx$ in terms of $sin(x)$, $cos(x)$, and $x$ which can be solved. I will update my answer. – unseen_rider Jul 18 '16 at 09:49
-
@unseen_rider I just can't see if, so if you can show it in more detail, even if not completely detailed, perhaps that will help. – DonAntonio Jul 18 '16 at 10:01
-
@michaelhardy please read updated answer and reconsider votes and explanations. – unseen_rider Jul 18 '16 at 14:19
-
@donantonio - please revisit my answer following additional information – unseen_rider Jul 18 '16 at 14:20
-
Sorry, still can't see it. You wrote in the first line after the $;u(x),v(x);$ thing, that $$\int_0^\infty \frac{cos(x)}{x^2}dx=-[cos(x)(\frac{-1}{x}]_0^\infty - \int_0^\infty \frac{sin(x)}{x}dx$$ but I just don't succeed in seeing it: if $;u=\cos x, v'=\frac1{x^2};$ , then the first term is $;uv'=-\frac{\cos x}{x^2};$ , and you have there a double minus sign... But besides this: the expression $$\left.-\frac{\cos x}x\right|_0^\infty$$ requires taking limits, and the limit at $;x=0;$ of this doesn't exis finitely... – DonAntonio Jul 18 '16 at 16:20
-
-
@unseen_rider : Can you write $\cos(x)$ instead of $cos(x)$? $\qquad$ – Michael Hardy Jul 18 '16 at 17:56
-