By Part Takes Forever!

Calculus Level 2

0 sin x x e x d x = ? \large \int_0^\infty \frac {\sin x}{xe^x} \, dx = \, ?

π \pi π 2 \frac{\pi}2 π 3 \frac{\pi}3 π 4 \frac \pi 4

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

8 solutions

Arjen Vreugdenhil
Aug 24, 2016

Lazy solution

Because sin x < x \sin x < x for x > 0 x > 0 0 sin x x e x < 0 x x e x = 0 e x = 1. \int_0^\infty \frac{\sin x}{xe^x} < \int_0^\infty \frac{x}{xe^x} = \int_0^\infty e^{-x} = 1. Of the four choices given, only π / 4 \pi/4 is less than one.

Epic trick for jee

subh mandal - 4 years, 6 months ago

What if Π/6 was also aviable

Satyam Tripathi - 4 years, 7 months ago

Log in to reply

Then more work is needed. But when faced with a multiple-choice question, I like to do no more work than is needed to rule out all incorrect answers.

Arjen Vreugdenhil - 4 years, 7 months ago

Yaa its a reasoning part but we can't rely upon these as we can face integer type ...BTW are u preparing for jee??

Satyam Tripathi - 4 years, 7 months ago

Log in to reply

No, it's been decades since I did any competing. I am a teacher in science who enjoys a good problem every now and then.

Arjen Vreugdenhil - 4 years, 7 months ago
Chew-Seong Cheong
Aug 17, 2016

Relevant wiki: Differentiation Under the Integral Sign

Let I ( a ) I(a) be as follows:

I ( a ) = 0 e a x sin x x d x I ( a ) a = 0 a e a x sin x x d x = 0 x e a x sin x x d x = 0 e a x sin x d x See note. = e a x ( a sin x + cos x ) 1 + a 2 0 = 1 1 + a 2 I ( a ) = 1 1 + a 2 d a = tan 1 a + C C is the constant of integration lim a I ( a ) = 0 lim a tan 1 a + C = 0 π 2 + C = 0 C = π 2 I ( 1 ) = tan 1 1 + π 2 = π 4 \begin{aligned} I(a) & = \int_0^\infty \frac {e^{-ax} \sin x}x dx \\ \frac {\partial I(a)}{\partial a} & = \int_0^\infty \frac {\partial}{\partial a} \frac {e^{-ax} \sin x}x dx \\ & = \int_0^\infty \frac {-xe^{-ax} \sin x}x dx \\ & = - \int_0^\infty e^{-ax} \sin x \ dx & \small \color{#3D99F6} \text{See note.} \\ & = \frac {e^{-ax}(a \sin x + \cos x)}{1+a^2} \bigg|_0^\infty \\ & = - \frac 1{1+a^2} \\ \implies I(a) & = - \int \frac 1{1+a^2} da \\ & = - \tan^{-1} a + \color{#3D99F6}{C} & \small \color{#3D99F6}{C \text{ is the constant of integration}} \\ \lim_{a \to \infty} I(a) & = 0 \\ \implies \lim_{a \to \infty} - \tan^{-1} a + C & = 0 \\ - \frac \pi 2 + C & = 0 \\ C & = \frac \pi 2 \\ \implies I(1) & = - \tan^{-1} 1 + \frac \pi 2 \\ & = \boxed{\dfrac \pi 4} \end{aligned}


Note:

J = 0 e a x sin x d x By integration by parts = e a x sin x a 0 + 0 e a x cos x a = e a x sin x a 0 + e a x cos x a 2 0 + 0 e a x sin x a 2 = e a x ( a sin x + cos x ) a 2 0 J a 2 a 2 J = e a x ( a sin x + cos x ) 0 J J = e a x ( a sin x + cos x ) 1 + a 2 0 \begin{aligned} J & = - \int_0^\infty e^{-ax} \sin x \ dx & \small \color{#3D99F6} \text{By integration by parts} \\ & = \frac {e^{-ax}\sin x}a \bigg|_0^\infty + \int_0^\infty \frac {e^{-ax}\cos x}a \\ & = \frac {e^{-ax}\sin x}a \bigg|_0^\infty + \frac {e^{-ax}\cos x}{a^2} \bigg|_0 ^\infty + \int_0^\infty \frac {e^{-ax}\sin x}{a^2} \\ & = \frac {e^{-ax}(a\sin x + \cos x)}{a^2} \bigg|_0 ^\infty - \frac J{a^2} \\ \implies a^2J & = e^{-ax}(a\sin x + \cos x) \bigg|_0 ^\infty - J \\ \implies J & = \frac {e^{-ax}(a\sin x + \cos x)}{1+a^2} \bigg|_0 ^\infty \end{aligned}

Moderator note:

Yes, this problem is a wonderful illustration on the importance of differentiation under the integral sign . Students might notice that if they tried to tackle this integral using by parts , they would end up with more monstrous integrals to deal with.

Food for thought:
Why did Chew-Seong start with I ( a ) = 0 sin a x x e x d x \displaystyle I(a)= \int_0^\infty \dfrac{\sin ax}{x e^x} \, dx , instead of say, I ( a ) = 0 sin x x a e x d x \displaystyle I(a)= \int_0^\infty \dfrac{\sin x}{x^a e^x} \, dx or I ( a ) = 0 sin x x e a x d x ? \displaystyle I(a)= \int_0^\infty \dfrac{\sin x}{x e^{ax}} \, dx?

Bonus question:
Let Ei ( ) \text{Ei}(\cdot) be a special exponential integral function: Ei = x e t t d t \displaystyle \text{Ei}= - \int_{-x}^\infty \frac{e^{-t}}t \, dt .
Then prove that we can rewrite the indefinite integral sin x x e x d x \displaystyle \int \displaystyle \dfrac{\sin x}{x e^x} \, dx in terms of Ei \text{Ei} .

Hint: e i x = cos x + i sin x e^{ix} = \cos x + i \sin x .

How are you justifying reversing the order of summation and integration in line 4? You cannot use the Monotone Convergence Theorem, since the series is alternating. Nor can you use the Dominated Convergence Theorem, since n 1 2 n + 1 \sum_n \tfrac{1}{2n+1} diverges.

Mark Hennings - 4 years, 10 months ago

Log in to reply

Don't quite understand what you mentioned. But it is 1 ( 2 n + 1 ) ! \sum \frac 1{(2n+1)!} which converges.

Chew-Seong Cheong - 4 years, 10 months ago

Log in to reply

No. The integral I n = 0 x 2 n e x ( 2 n + 1 ) ! d x = 1 2 n + 1 I_n \;= \; \int_0^\infty \frac{x^{2n}e^{-x}}{(2n+1)!}\,dx \; = \; \frac{1}{2n+1} and you are wanting to sum n = 0 ( 1 ) n I n \sum_{n=0}^\infty (-1)^n I_n in line 4. To use the DCT to reverse the order of summation and integration (obtaining an equation between lines 3 and 4), you would need something like the convergence of n = 0 0 ( 1 ) n x 2 n e x ( 2 n + 1 ) ! d x = n = 0 I n \sum_{n=0}^\infty \int_0^\infty \left|(-1)^n\frac{x^{2n}e^{-x}}{(2n+1)!}\,dx\right| \; = \; \sum_{n=0}^\infty I_n which does not exist.

Although it is legitimate to interchange the integral and the limit (you do get the right answer), the reason for being able to do so needs a little work...

Mark Hennings - 4 years, 10 months ago

Log in to reply

@Mark Hennings Thanks, I got it.

Chew-Seong Cheong - 4 years, 10 months ago

Plz will u rewrite that solution again bcoz it is not clearly visible

vishwesh agrawal - 4 years, 9 months ago

Log in to reply

My discussion with Mark Hennings was for a wrong solution I posted earlier. The one shown here is correct and not related to the discussion.

Chew-Seong Cheong - 4 years, 9 months ago

"By Part Takes Forever!" The correct terminology here would be "Integration by parts", as several respondents wrote correctly. I don't care what part of the world you are from, you need to use the right mathematical terminology if you are writing in English. I would also suggest to try solving this problem by using complex variables, contour integration, and the residue theorem in complex analysis (developed by the Frenchman Cauchy, as did much of this subject).

Dale Wood - 4 years, 9 months ago

Log in to reply

If using complex variables works, it would entail drawing diagrams to show what the right contour is and to show where the poles are inside the contour and on it. Clever use of limits as the radii of parts of the contour approach infinity or zero is also required. Here is not a good place to draw the diagrams. I love complex analysis because it is quite visual.

Dale Wood - 4 years, 9 months ago

Log in to reply

I studied complex analysis in two different courses in graduate school. In the first one (the larger one) we had several Chinese students who often discussed things in Chinese. However, in the midst of this, they said "removable singularity" in English in the midst of the Chinese. I asked them why, and one man explained to me that they did not know the Chinese expression for "removable singularity", so they just used the English phrase. I loved this answer because I like learning about other cultures, and learning of the difficulties of going to college in the USA, Canada, England, etc., when your native language is Chinese, Spanish, Arabic, Hindi, Vietnamese, etc.

Dale Wood - 4 years, 9 months ago

can you plz explain how you wrote 5th step from the forth one ? plz i am not able to understand this fenyman trick ! plz help sir !

A Former Brilliant Member - 4 years, 5 months ago

Log in to reply

I have added a note. Hope it is useful.

Chew-Seong Cheong - 4 years, 5 months ago

Log in to reply

thank you very much for explaining sir, you rock !!

A Former Brilliant Member - 4 years, 5 months ago

Can you explain why does we take the 'limit' of I(a)? @Chew-Seong Cheong

Muhammad Alif - 2 years, 7 months ago

Log in to reply

We need to have a value for I ( a ) I(a) , so that we can find the value of C C .

Chew-Seong Cheong - 2 years, 7 months ago
Refaat M. Sayed
Aug 17, 2016

Another solution using laplace transform 0 sin ( t ) t e t d t = lim s 0 0 e s t e t sin ( t ) t d t = lim s 0 ( e t sin ( t ) t ) = lim s 0 ( π 2 arctan ( s + 1 ) ) = π 2 π 4 = π 4 \int \limits^{\infty }_{0}\frac{\sin \left( t\right) }{te^{t}} dt \\ =\lim \limits_{s\to 0}\int \limits^{\infty }_{0}\frac{e^{-st}e^{-t}\sin \left( t\right) }{t} dt \\ =\lim \limits_{s\to 0}\ell \left( \frac{e^{-t}\sin \left( t\right) }{t} \right) \\ =\lim \limits_{s\to 0}\left( \frac{\pi }{2} -\arctan \left( s+1\right) \right) \\ =\frac{\pi }{2} -\frac{\pi }{4} =\boxed{\frac{\pi }{4}}

That's a good approach, using the Laplace Transform, named for the Frenchman Pierre-Simon Laplace and hence capitalized. It also shows the connection of this problem with complex analysis because taking inverse Laplace Transforms usually involves (in the nitty-gritty) using complex variables, contour integration, and the residue theorem. Undergraduate students get to bypass this because they (and we) can use tables of Laplace Transform pairs to bypass all of that. This works because the Laplace transform and the inverse transform between f(t) and its transform F(s) are unique. In other words, "it goes both ways".

Dale Wood - 4 years, 9 months ago

Log in to reply

Here is a good way to remember how terminology goes in mathematics. With just one exception, ALL of the theorems, axioms, lemmas, procedures, functions, equations, and densities that are named for PEOPLE are always capitalized. The lone exception is "abelian", named for Niels Henrik Abel, and abelian got this way just because it is so common. For examples of the contrary, there are Cauchy's residue theorem, the Cauchy-Riemann equations, several Euler theorems and functions, the Fourier transform, the Gaussian density, Hamiltonians, the Jacobean, the Laplace transform, Lesbegue integration, Newton's procedure, the Pascal triangle, the Schwartz inequality, the Tchebyshev inequality, and the Von Neumann procedure in computers.

Dale Wood - 4 years, 9 months ago

Log in to reply

Other important items in mathematics are named for the Bernoullis, Carmichael, Gödel, Heaviside, Jordan, Lagrange, Leibnitz, Markov, Maclauren, Rayleigh, and Taylor. This goes to show you that the best procedure is to capitalize ALL of these names, and we would forgive you for "Abelian". There is even something, very useful in physics, that could be called the Lorentz - Fitzgerald - Einstein transform.

Dale Wood - 4 years, 9 months ago

Relevant; Leibniz Formula for π \pi

By Taylor expansion we know that

sin x = n = 0 ( 1 ) n x 2 n + 1 ( 2 n + 1 ) ! = x x 3 3 ! + x 5 5 ! \begin{aligned} \sin x= \sum_{n=0}^\infty \frac{(-1)^n x^{2n+1} }{(2n+1)!}=x-\frac{x^3}{3!}+\frac{x^5}{5!}-\cdots \end{aligned}

which implies

0 e x sin x x = 0 n = 0 e x ( 1 ) n x 2 n ( 2 n + 1 ) ! \begin{aligned} \int_0^{\infty} e^{-x} \frac{\sin x}{x} &= \int_0^{\infty} \sum_{n=0}^\infty e^{-x} \frac{(-1)^n x^{2n}}{(2n+1)!} \end{aligned}

It is easy to see (using integration by parts and induction) that for m N m \in \mathbb{N}

0 e x x m = m ! \begin{aligned} \int_0^\infty e^{-x} x^m=m! \end{aligned}

Given that the series converges (by the ratio test) to an integrable limit we can change the order of integration using Fubini's Theorem

0 n = 0 e x ( 1 ) n x 2 n ( 2 n + 1 ) ! = n = 0 0 e x ( 1 ) n x 2 n ( 2 n + 1 ) ! = 0 e x ( 1 x 2 3 ! + x 4 5 ! = 1 1 3 + 1 5 = π 4 , \begin{aligned} \int_0^{\infty} \sum_{n=0}^\infty e^{-x} \frac{(-1)^n x^{2n}}{(2n+1)!} &= \sum_{n=0}^\infty \int_0^{\infty} e^{-x} \frac{(-1)^n x^{2n}}{(2n+1)!}\\ &= \int_0^{\infty} e^{-x} (1-\frac{x^2}{3!}+\frac{x^4}{5!}-\dots\\ &=1-\frac{1}{3}+\frac{1}{5}-\dots =\frac{\pi}{4}, \end{aligned}

Where the last equality holds by the Leibniz formula for π \pi .

Precisely the same solution as mine

Ceesay Muhammed - 4 years, 7 months ago
Aaryan Jain
Aug 17, 2016

consider I(a)= s i n ( a x ) x e x \frac{sin(ax)}{xe^x} now integral of this withrespect to dx will be a fn of "a" partial differentiatin wrt "a" I'(a) =

now apply by parts.. and then get I(a) a=1 at last

Daniel Ellesar
Dec 15, 2020

This solution uses methods of complex analysis related to Cauchy's Residue Theorem. In particular, we make use of Cauchy's Theorem.

Let I = 0 sin x x e x d x I=\int_0^\infty \frac{\sin x}{x e^x}dx . Then let f + ( z ) = 1 z e ( i 1 ) z f_+(z)=\frac{1}{z}e^{(i-1)z} and f ( z ) = 1 z e ( i 1 ) z f_-(z)=\frac{1}{z}e^{(-i-1)z} .

Observe that I = I m ( 0 f + ( z ) d z ) = I m ( 0 f ( z ) d z ) I=\mathcal{Im}(\int_0^\infty f_+ (z)dz)=-\mathcal{Im}(\int_0^\infty f_- (z)dz) .

Consider for ϵ , R > 0 \epsilon, R>0 , γ \gamma the contour in the complex plane consisting of:

  • γ ϵ \gamma_\epsilon , a quarter arc of radius ϵ \epsilon , centred at zero,

  • γ R \gamma_R , a quarter arc of radius R R , centred at zero,

  • and the real and imaginary line segments connecting them,

the whole contour in positive orientation.

Then it follows that: γ R f + ( z ) d z + γ ϵ f + ( z ) d z + ϵ R f + ( z ) d z + i R i ϵ f + ( z ) d z = γ f + ( z ) d z \int_{\gamma_R} f_+ (z)dz+\int_{\gamma_\epsilon} f_+ (z)dz+\int_\epsilon^R f_+ (z) dz+\int_{iR}^{i\epsilon} f_+ (z) dz = \int_{\gamma} f_+ (z)dz .

But γ f + ( z ) d z = 0 \int_{\gamma} f_+ (z)dz = 0 as f + f_+ is holomorphic on γ \gamma , and I m ( ϵ R f + ( z ) d z ) = I m ( i R i ϵ f + ( z ) d z ) = I \mathcal{Im}(\int_\epsilon^R f_+ (z) dz)=\mathcal{Im}(\int_{iR}^{i\epsilon} f_+ (z) dz)=I as ϵ 0 , R \epsilon\to 0, R\to \infty , which we can see by applying the substitution z = i x z=ix to the second integral.

Hence we have I m ( γ R f + ( z ) d z ) + I m ( γ ϵ f + d z ) + 2 I = 0 \mathcal{Im}(\int_{\gamma_R} f_+ (z)dz)+\mathcal{Im}(\int_{\gamma_\epsilon} f_+dz) +2I=0 (as ϵ 0 , R \epsilon\to 0, R\to \infty )

It remains to find lim R γ R f + ( z ) d z \lim_{R\to\infty} \int_{\gamma_R} f_+ (z)dz and lim ϵ 0 γ ϵ f + ( z ) d z \lim_{\epsilon\to 0} \int_{\gamma_\epsilon} f_+ (z)dz .

Indeed we have γ R f + ( z ) d z π 2 R f + π 2 e R ||\int_{\gamma_R} f_+ (z)dz||\leq \frac{\pi}{2} R ||f_+ || \leq \frac{\pi}{2e^R} as e i z ||e^{iz}|| is bounded above by one on the upper half plane. Hence as R R\to\infty , γ R f + ( z ) d z 0 \int_{\gamma_R} f_+ (z)dz\to 0 .

Now we apply the substitution z = ϵ e i t : 0 t < π 2 z=\epsilon e^{it}:0\leq t < \frac{\pi}{2} to get γ ϵ f + ( z ) d z = π 2 0 1 ϵ e i t e ϵ ( i 1 ) e i t × ϵ i e i t d t = i 0 π 2 e ϵ ( i 1 ) e i t d t ( ) \int_{\gamma_\epsilon} f_+ (z)dz=\int_{\frac{\pi}{2}}^0 \frac{1}{\epsilon e^{it}} e^{\epsilon (i-1)e^{it}}\times \epsilon i e^{it}dt=-i\int_0^{\frac{\pi}{2}} e^{\epsilon (i-1)e^{it}} dt (*) .

Taking ϵ 0 \epsilon \to 0 we get ( ) i 0 π 2 d t = i π 2 (*)\to -i\int_0^{\frac{\pi}{2}} dt=-i\frac{\pi}{2} .

Putting it all together: I m ( 0 ) + I m ( i π 2 ) + 2 I = 0 I = π 4 \mathcal{Im}(0) + \mathcal{Im}(-i\frac{\pi}{2}) + 2I = 0\implies I=\boxed{\frac{\pi}{4}}

Ameya Anjarlekar
Aug 25, 2016

Max value of the function is 1 and it is decreasing so area must be <1 opt D

Qwerty Asdfghjkl
May 7, 2018

Just use diff under imtegral sign by replacing sinx by sin ax

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...