Integration

Calculus Level 4

I ( n ) = 0 π ln ( 1 2 n cos x + n 2 ) d x \large I(n) = \int_0^\pi \ln (1-2n\cos x + n^2) \, dx

For I ( n ) I(n) as defined above, find I ( 100 ) I ( 10 ) I ( 36 ) I ( 6 ) \dfrac {I(100)}{I(10)}\cdot \dfrac {I(36)}{I(6)} .


The answer is 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.

4 solutions

Guilherme Niedu
Apr 20, 2017

I ( n ) = 0 π ln ( 1 2 n cos ( x ) + n 2 ) d x \large \displaystyle I(n) = \int_0^{\pi} \ln(1 - 2n \cos(x) + n^2) dx

Since a b f ( x ) d x = a b f ( a + b x ) d x \color{#20A900} \int_a^b f(x)dx = \int_a^b f(a+b-x)dx :

I ( n ) = 1 2 [ 0 π ln ( 1 2 n cos ( x ) + n 2 ) d x + 0 π ln ( 1 2 n cos ( 0 + π x ) + n 2 ) d x ] \large \displaystyle I(n) = \frac12 \left [ \int_0^{\pi} \ln(1 - 2n \cos(x) + n^2) dx + \int_0^{\pi} \ln(1 - 2n \cos(0 + \pi - x) + n^2) dx \right ]

I ( n ) = 1 2 [ 0 π ln ( 1 2 n cos ( x ) + n 2 ) d x + 0 π ln ( 1 + 2 n cos ( x ) + n 2 ) d x ] \large \displaystyle I(n) = \frac12 \left [ \int_0^{\pi} \ln(1 - 2n \cos(x) + n^2) dx + \int_0^{\pi} \ln(1 + 2n \cos(x) + n^2) dx \right ]

I ( n ) = 1 2 [ 0 π ln ( 1 + 2 n 2 + n 4 4 n 2 cos 2 ( x ) ) d x ] \large \displaystyle I(n) = \frac12 \left [ \int_0^{\pi} \ln (1 + 2n^2 + n^4 - 4n^2 \cos^2(x) ) dx \right ]

I ( n ) = 1 2 [ 0 π ln ( 1 + 2 n 2 ( 1 2 cos 2 ( x ) ) + n 4 ) d x ] \large \displaystyle I(n) = \frac12 \left [ \int_0^{\pi} \ln (1 + 2n^2 (1- 2 \cos^2(x)) + n^4 ) dx \right ]

I ( n ) = 1 2 [ 0 π ln ( 1 2 n 2 cos ( 2 x ) + n 4 ) d x ] \large \displaystyle I(n) = \frac12 \left [ \int_0^{\pi} \ln (1 - 2n^2 \cos(2x) + n^4 ) dx \right ]

u = 2 x d x = 1 2 d u \color{#20A900} \large \displaystyle u = 2x \rightarrow dx = \frac12 du

I ( n ) = 1 4 [ 0 2 π ln ( 1 2 n 2 cos ( u ) + n 4 ) d u ] \large \displaystyle I(n) = \frac14 \left [ \int_0^{2 \pi} \ln (1 - 2n^2 \cos(u) + n^4 ) du \right ]

I ( n ) = 1 4 [ 0 π ln ( 1 2 n 2 cos ( u ) + n 4 ) d u + π 2 π ln ( 1 2 n 2 cos ( u ) + n 4 ) d u ] \large \displaystyle I(n) = \frac14 \left [ \int_0^{\pi} \ln (1 - 2n^2 \cos(u) + n^4 ) du + \int_{\pi}^{2 \pi} \ln (1 - 2n^2 \cos(u) + n^4 ) du \right ]

On second integral:

v = 2 π u d u = d v \color{#20A900} \large \displaystyle v = 2 \pi - u \rightarrow du = - dv

I ( n ) = 1 4 [ 0 π ln ( 1 2 n 2 cos ( u ) + n 4 ) d u π 0 ln ( 1 2 n 2 cos ( v ) + n 4 ) d v ] \large \displaystyle I(n) = \frac14 \left [ \int_0^{\pi} \ln (1 - 2n^2 \cos(u) + n^4 ) du - \int_{\pi}^{0} \ln (1 - 2n^2 \cos(v) + n^4 ) dv \right ]

I ( n ) = 1 4 [ 0 π ln ( 1 2 n 2 cos ( u ) + n 4 ) d u + 0 π ln ( 1 2 n 2 cos ( v ) + n 4 ) d v ] \large \displaystyle I(n) = \frac14 \left [ \int_0^{\pi} \ln (1 - 2n^2 \cos(u) + n^4 ) du + \int_{0}^{\pi} \ln (1 - 2n^2 \cos(v) + n^4 ) dv \right ]

I ( n ) = 1 4 [ I ( n 2 ) + I ( n 2 ) ] \large \displaystyle I(n) = \frac14 \left [ I(n^2) + I(n^2) \right ]

I ( n ) = 1 2 I ( n 2 ) \large \displaystyle I(n) = \frac12 I(n^2)

I ( n 2 ) I ( n ) = 2 \color{#20A900} \boxed { \large \displaystyle \frac{I(n^2)}{I(n)} = 2 }

So:

I ( 100 ) I ( 10 ) I ( 36 ) I ( 6 ) = I ( 1 0 2 ) I ( 10 ) I ( 6 2 ) I ( 6 ) = 2 2 = 4 \color{#3D99F6} \large \displaystyle \frac{I(100)}{I(10)} \cdot \frac{I(36)}{I(6)} = \frac{I(10^2)}{I(10)} \cdot \frac{I(6^2)}{I(6)} = 2\cdot 2 = \boxed {\large \displaystyle 4}

There is a mistake in fifth line.It will be cos^2(x) not cos(x).

Nice one (+1)

Kushal Bose - 4 years, 1 month ago

Log in to reply

Thanks! Fixed it

Guilherme Niedu - 4 years, 1 month ago

Cant it be solved by simplifying the expression inside I(n) to ln[(1+ne^{ix})(1-ne^{-ix})] and using taylor series for ln(x)

Abhash Jha - 4 years, 1 month ago

Log in to reply

sorry it is (1+e^{-ix}) not (1-e^{-ix}

Abhash Jha - 4 years, 1 month ago
Mark Hennings
Apr 20, 2017

We can evaluate the integrals, rather than just find relationships between them. Note that for μ > 0 \mu > 0 , F ( μ ) = 0 π sin μ 1 x d x = 2 μ 1 0 π sin μ 1 1 2 x cos μ 1 1 2 x d x = 2 μ 0 1 2 π sin μ 1 x cos μ 1 x d x = 2 μ 1 B ( 1 2 μ , 1 2 μ ) = 2 μ 1 Γ ( 1 2 μ ) 2 Γ ( μ ) \begin{aligned} F(\mu) & = \int_0^\pi \sin^{\mu-1}x\,dx \; = \; 2^{\mu-1}\int_0^\pi \sin^{\mu-1}\tfrac12x \cos^{\mu-1}\tfrac12x\,dx \; = \; 2^\mu \int_0^{\frac12\pi} \sin^{\mu-1}x \cos^{\mu-1}x\,dx \\ & = 2^{\mu-1}B\big(\tfrac12\mu,\tfrac12\mu\big) \; = \; \frac{2^{\mu-1}\Gamma(\tfrac12\mu)^2}{\Gamma(\mu)} \end{aligned} so that F ( μ ) = F ( μ ) [ ln 2 + ψ ( 1 2 μ ) ψ ( μ ) ] F'(\mu) \; = \; F(\mu)\big[\ln2 + \psi\big(\tfrac12\mu\big) - \psi(\mu)\big] and hence 0 π ln ( sin x ) d x = F ( 1 ) = F ( 1 ) [ ln 2 + ψ ( 1 2 ) ψ ( 1 ) ] = π [ ln 2 γ ln 4 + γ ] = π ln 2 \int_0^\pi \ln(\sin x)\,dx \; = \; F'(1) \; = \; F(1)\big[\ln2 + \psi(\tfrac12) - \psi(1)\big] \; = \; \pi\big[\ln2 - \gamma - \ln4 + \gamma\big] \; = \; -\pi\ln2 and hence we obtain the moderately standard integral 0 π ln ( 2 sin x ) d x = 0 \int_0^\pi \ln(2\sin x)\,dx \; = \; 0 If we now define I ( α ) = 0 π ln ( 1 2 α cos x + α 2 ) d x = 1 2 0 2 π ln ( 1 2 α cos x + α 2 ) d x α 1 I(\alpha) \; = \; \int_0^\pi \ln(1 - 2\alpha \cos x + \alpha^2)\,dx \; = \; \tfrac12\int_0^{2\pi} \ln(1 - 2\alpha \cos x + \alpha^2)\,dx \hspace{1cm} \alpha \ge 1 then I ( 1 ) = 1 2 0 2 π ln ( 4 sin 2 1 2 x ) d x = 0 2 π ln ( 2 sin 1 2 x ) d x = 2 0 π ln ( 2 sin x ) d x = 0 I(1) \; = \; \tfrac12\int_0^{2\pi}\ln\big(4\sin^2\tfrac12x\big)\,dx \; = \; \int_0^{2\pi} \ln\big(2\sin\tfrac12x\big)\,dx \; = \; 2\int_0^\pi \ln(2\sin x)\,dx \; = \; 0 Moreover the substitution z = e i x z = e^{ix} and some contour integration yields I ( α ) = 1 2 0 2 π 2 α 2 cos x 1 2 α cos x + α 2 = 1 2 z = 1 2 α z z 1 1 α ( z + z 1 ) + α 2 d z i z = 1 2 i z = 1 2 α z z 2 1 z ( z α ) ( α z 1 ) d z = π ( R e s z = 0 + R e s z = α 1 ) 2 α z z 2 1 z ( z α ) ( α z 1 ) = π ( 1 α + 1 α 2 α 1 α ) = 2 π α \begin{aligned} I'(\alpha) & = \tfrac12\int_0^{2\pi} \frac{2\alpha - 2\cos x}{1 - 2\alpha \cos x + \alpha^2} \; = \; \tfrac12 \int_{|z|=1} \frac{2\alpha - z - z^{-1}}{1 - \alpha(z + z^{-1}) + \alpha^2}\,\frac{dz}{i z} \\ & = -\frac{1}{2i}\int_{|z|=1} \frac{2\alpha z - z^2 - 1}{z(z-\alpha)(\alpha z - 1)}\,dz \; = \; -\pi\Big(\mathrm{Res}_{z=0} + \mathrm{Res}_{z=\alpha^{-1}}\Big) \frac{2\alpha z - z^2 - 1}{z(z-\alpha)(\alpha z - 1)} \\ & = -\pi\Big(-\frac{1}{\alpha} + \frac{1 - \alpha^{-2}}{\alpha^{-1} - \alpha}\Big) \; = \; \frac{2\pi}{\alpha} \end{aligned} for α > 1 \alpha > 1 , so that I ( α ) = I ( 1 ) + 2 π ln α = 2 π ln α α 1 I(\alpha) \; = \; I(1) + 2\pi\ln\alpha \; = \; 2\pi\ln\alpha \hspace{2cm} \alpha \ge 1 so that I ( α 2 ) I ( α ) = 2 α 1 \frac{I(\alpha^2)}{I(\alpha)} \; =\; 2 \hspace{2cm} \alpha \ge 1 making the answer to the question equal to 4 \boxed{4} .

Wow!

Thanks a lot for sharing this with us, sir!

Rohith M.Athreya - 4 years, 1 month ago

Very nice! Powerful integration technique!

Wei Chen - 4 years, 1 month ago

Log in to reply

It is also possible to do this by considering the electrostatic potential of an infinite charged cylindrical shell of radius 1 1 with constant surface charge density σ \sigma . The potential at a distance r r from the axis of the cylinder is then σ 2 π ε 0 I ( r ) -\frac{\sigma}{2\pi\varepsilon_0}I(r) for r > 1 r > 1 . On the other hand, it is clear that the electrostatic potential of this cylindrical shell will only depend on this radial distance r r , and hence (for r > 1 r > 1 ) will be the same as that of an infinite charged line with linear charge density 2 π σ 2\pi\sigma positioned along the axis of the cylinder. Thus σ 2 π ε 0 I ( r ) = 2 π σ 2 π ε 0 ln r r > 1 -\frac{\sigma}{2\pi\varepsilon_0}I(r) \; = \; -\frac{2\pi \sigma}{2\pi \varepsilon_0} \ln r \hspace{1cm} r > 1 and hence it follows that I ( r ) = 2 π ln r I(r) = 2\pi \ln r for r > 1 r > 1 .

Mark Hennings - 4 years, 1 month ago

Log in to reply

Yeah, that's another nice way to look at it. To get LHS we integrate over the azimuth angle, and to get to RHS we can simply apply Gauss's law. Very neat!

Wei Chen - 4 years, 1 month ago

this is new and so cool

avi solanki - 4 years, 1 month ago
Prakhar Bindal
Apr 20, 2017

I'm a bit busy these days . this problem is from AIITS 2 Of FIITJEE.

Try proving I(n^2)/I(n) = 2 by using a+b-x property and adding

Just saw your suggestion, sir. I did the same approach to answer it, it's below :)

Guilherme Niedu - 4 years, 1 month ago

Log in to reply

Please you are older than me . call me prakhar! :) Nice solution BTW

Prakhar Bindal - 4 years, 1 month ago

Log in to reply

Thanks, Prakhar!

Guilherme Niedu - 4 years, 1 month ago

This integral is popularly known as Dini's integral(in honour of Italian Mathematician Ulisse Dini) and takes the values 0 0 for 0 < = n < 1 0<=n<1 .And 2 π l n ( n ) 2πln(n) for n > 1 n>1

Spandan Senapati - 4 years, 1 month ago
Kaustubh Mishra
May 9, 2017

You may also approach by writting (1-2ncos(x) + n 2 n^{2} ) = (n- e x e^{x} )(n- e x e^{-x} )

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...