0

Find whether $\int_{1}^{\infty} \frac{sin(x)}{x} dx$ is converging or not.

I tried to use comparison test or limit comparison test but could't find a suitable function. How can I determine what type of a function to use when using these tests?

S.Dan
  • 1,115
  • Integration by parts ! ;) – ParaH2 Jun 03 '15 at 11:59
  • Almost duplicate... http://math.stackexchange.com/questions/5248/solving-the-integral-int-0-infty-frac-sinxx-dx-frac-pi2 – JP McCarthy Jun 03 '15 at 12:00
  • 2
    @JpMcCarthy Determining if the integral converges is not the same thing as evaluating the integral. – Gappy Hilmore Jun 03 '15 at 12:08
  • 1
    @grdgfgr but if you solve it and it converges to a particular value... – snulty Jun 03 '15 at 12:09
  • a) I said it was an almost duplicate and I said this because b) many of the answers therein talk about convergence... just do CTRL-F "converge"... also what snulty said. – JP McCarthy Jun 03 '15 at 12:09
  • You can sum all positive integers and obtain $-1/12$, doesnt mean the series converges. – Gappy Hilmore Jun 03 '15 at 12:10
  • This was linked on the right hand side- http://math.stackexchange.com/questions/1089495/convergence-of-int-1-infty-frac-sin-xx-alphadx?rq=1 – snulty Jun 03 '15 at 12:11
  • @grdgfgr The answer above this addresses the question http://math.stackexchange.com/questions/5248/solving-the-integral-int-0-infty-frac-sinxx-dx-frac-pi2#comment325673_106434 – JP McCarthy Jun 03 '15 at 12:12
  • @grdgfgr by the way I like your answer and +1'd it. – JP McCarthy Jun 03 '15 at 12:13
  • @snulty Let's assume $S=\sum_{n=0}^\infty 2^n$ converged. We would multiply it by 2, subtract itself from it and observe that $2S-S=-1$ This demonstrates that obtaining a finite answer by evaluating an integral/summation gives no information on whether or not it actually converges. You cannot assume convergence, integrate/sum, and then decide if it converged or not. – Gappy Hilmore Jun 03 '15 at 12:31
  • @snulty Who knows, perhaps $\int sinx,/ x , dx$ infact diverged, but while trying to evaluate the integral, we performed unjustified manipulations which led us to a finite answer. And this happens all the time if you try to evaluate divergent integrals. Oftentimes, you will get a finite answer. – Gappy Hilmore Jun 03 '15 at 12:32
  • http://en.wikipedia.org/wiki/Dirichlet%27s_test – Jack D'Aurizio Jun 03 '15 at 14:50

3 Answers3

4

Consider the area of each lobe of the function to be the sequence $a_n$

The area alternates in sign, and due to $1/x$, it decreases monotonically.

By alternating series test, it will converge.

1

This, like the analogous series $\sum \limits _{n=1} ^\infty \frac {\sin n} n$, is the prototypical application of the Abel-Dirichlet test. It's a subtle problem, no other convergence test among the usual ones would do the job. Please note that if the question had been about $\int \limits _1 ^\infty \frac {\sin x} {x^p} \Bbb d x$ with $p > 1$, then it would have been much easier.

The Abel-Dirichlet test tells us that if we have two functions $\alpha, f : [a,\infty) \to [0,\infty)$ such that $\alpha$ decreases to $0$ and there exist $M>0$ with $|\int \limits _a ^b f (x) \Bbb d x| \leq 0 \space \forall b \geq a$, then $\int \limits _a ^\infty \alpha(x) f(x) \Bbb d x$ converges.

With the notations above, take $a=1, \space \alpha = \frac 1 x, \space f = \sin x$. That $\frac 1 x$ satisfies the conditions on $\alpha$ is trivial to check; concerning $f$, note that $|\int \limits _1 ^b \sin x \Bbb d x| = |- \cos b + \cos 1| \leq |\cos b| + |\cos 1| \leq 2$, so choose $M=2$ and Abel and Dirichlet do the rest for you.

Alex M.
  • 35,207
0

Bound it as following:

$$\left|\int_{n\pi}^{(n+1)\pi} \frac{\sin{x}}{x}dx\right|<\left|\frac{1}{x}\Bigg|_{n\pi}\int_{n\pi}^{(n+1)\pi}\sin x\,dx\right|=\frac1{n\pi}$$ $$\left|\int_{n\pi}^{(n+1)\pi} \frac{\sin{x}}{x}dx\right|>\left|\frac{1}{x}\Bigg|_{(n+1)\pi}\int_{n\pi}^{(n+1)\pi}\sin x\,dx\right|=\frac1{(n+1)\pi}$$

where I used that $1/x$ is monotonously decreasing (therefore the biggest on the lower bound.

EDIT:

The entire integral can be written as:

$$\int_1^\infty \frac{\sin x}{x}\,dx=-\int_0^1\frac{\sin x}{x}\,dx+\sum_{n=0}^\infty \int_{n\pi}^{(n+1)\pi}\frac{\sin x}{x}\,dx$$

For this specific choice of bounds, we can factor out the alternating sign and write: $$\sum_{n=0}^\infty \int_{n\pi}^{(n+1)\pi}\frac{\sin x}{x}\,dx=\sum_{n=0}^\infty (-1)^n\left|\int_{n\pi}^{(n+1)\pi}\frac{\sin x}{x}\,dx\right|$$ Now you have an alternating series, which is known to converge if the absolute value is monotonously decreasing. Which it is, because we bounded it very efficiently.

Comment: if you argue that $\int_0^1$ is also not proven to converge, just write the integral as $\int_1^\pi+\sum_{n=1}$ instead.

orion
  • 15,781
  • 1
    @AlexM., GPerez, and hamidkamali: $\left.\frac{1}{x}\right|_{n\pi}$ here means "the function $\frac{1}{x}$ evaluated at $n\pi$". It is not a typo. Moreover, your revisions are not mathematically correct. Please be more careful when reviewing and/or editing. – epimorphic Jun 03 '15 at 13:57
  • Could you continue your reasoning, please? If I am correct, this approach leads nowhere, i.e it will show that the given integral is less than a divergent series (that tends to infinity), which does not prove anything. – Alex M. Jun 03 '15 at 16:22