The Student Room Group

The Proof is Trivial!

Scroll to see replies

ImageUploadedByStudent Room1447889993.955461.jpg
Unless ive missed something terribly obvious 😁 i havent done integration in a while.. Since step III actually.


Posted from TSR Mobile
Original post by atsruser
Problem 546 (*/**)

Evaluate In=0π211+tann(x) dx\displaystyle I_n = \int_0^\frac{\pi}{2} \frac{1}{1+\tan^n(x)} \ dx

Solution 546:

In=0π211+tann(x) dx=0π2cosn(x)sinn(x)+cosn(x) dxI_n=\displaystyle\int_0^{\frac{ \pi}{2}} \dfrac{1}{1+\tan^n(x)} \ dx = \displaystyle\int_0^{\frac{\pi}{2}} \dfrac{\cos^n(x)}{\sin^n(x)+\cos^n(x)} \ dx
Sending xπ2xx \mapsto \dfrac{\pi}{2}-x yields:
In=0π2sinn(x)sinn(x)+cosn(x) dx[br]2In=0π21 dx[br]In=π4I_n=\displaystyle\int_0^{\frac{\pi}{2}} \dfrac{\sin^n(x)}{\sin^n(x)+\cos^n(x)} \ dx[br]\Rightarrow 2I_n=\displaystyle\int_0^{\frac{\pi}{2}} 1 \ dx[br]\Rightarrow I_n=\dfrac{\pi}{4}.

Oh snap, 2nd place :s-smilie:.
(edited 8 years ago)
I missed the obvious steps as its not like we all don't know what is going on.


Posted from TSR Mobile
Original post by joostan
Solution 546:

In=0π211+tann(x) dx=0π2cosn(x)sinn(x)+cosn(x) dxI_n=\displaystyle\int_0^{\frac{\pi}{2}} \dfrac{1}{1+\tan^n(x)} \ dx = \displaystyle\int_0^{\frac{\pi}{2}} \dfrac{\cos^n(x)}{\sin^n(x)+\cos^n(x)} \ dx
Sending xπ2xx \mapsto \dfrac{\pi}{2}-x yields:
In=0π2sinn(x)sinn(x)+cosn(x) dx[br]2In=0π21 dx[br]In=π4I_n=\displaystyle\int_0^{\frac{\pi}{2}} \dfrac{\sin^n(x)}{\sin^n(x)+\cos^n(x)} \ dx[br]\Rightarrow 2I_n=\displaystyle\int_0^{\frac{\pi}{2}} 1 \ dx[br]\Rightarrow I_n=\dfrac{\pi}{4}.

Thanks, someone did it in latex. I really need to learn how to use this..


Posted from TSR Mobile
Original post by physicsmaths
Thanks, someone did it in latex. I really need to learn how to use this..


Posted from TSR Mobile


It ain't all that hard :smile: although on TSR it can be quite buggy at times.
Original post by joostan
It ain't all that hard :smile: although on TSR it can be quite buggy at times.


Very true.

Also, love the solution. :biggrin:
Original post by physicsmaths
ImageUploadedByStudent Room1447889993.955461.jpg
Unless ive missed something terribly obvious 😁 i havent done integration in a while.. Since step III actually.


Posted from TSR Mobile


Yes, that's fine. It's very straightforward if you know what to do, but it's a pretty nice result anyway.
Original post by atsruser
Yes, that's fine. It's very straightforward if you know what to do, but it's a pretty nice result anyway.


Very pretty. :smile:
Original post by Indeterminate
Nicely done, and I don't blame you for that second residue :lol:


As I guess you know, that substitution is standard for integrals involving sin and cos between 0 and 2pi.

And as for the second residue, yes, I simply brute forced it rather than thinking about any cleaner approach. And speaking of cleaner approaches ..


Besides finding the second derivative, another way of calculating it would involve considering a small circle around the origin. The residue would then be the coefficient of z2z^2 in the expansion of

1(2z1)(z2)\dfrac{1}{(2z-1)(z-2)}

but that's hardly mess-free :s-smilie:


.. I'm not sure I'm following you. Where does the coefficient of z^2 come into the picture? I'm trying to imagine a Laurent series method here, but that doesn't seem to be what you're doing.
Original post by Zacken
Very pretty. :smile:


I guess an added bonus question would be: come up with an intuitive explanation for why all of those integrals yield the same result.

I don't have one, but it seems to me that there must be a more revealing way of getting to that answer, apart from just doing the integral.
Problem 547

Evaluate 0(sinxx)ndx\displaystyle \int_0^{\infty} \left(\frac{\sin x}{x}\right)^n \, \mathrm{d}x for n=1,2,3n=1, 2, 3

Spoiler

Spoiler

(edited 8 years ago)
Original post by atsruser
As I guess you know, that substitution is standard for integrals involving sin and cos between 0 and 2pi.

And as for the second residue, yes, I simply brute forced it rather than thinking about any cleaner approach. And speaking of cleaner approaches ..

.. I'm not sure I'm following you. Where does the coefficient of z^2 come into the picture? I'm trying to imagine a Laurent series method here, but that doesn't seem to be what you're doing.


Well here's my method:

Spoiler

(edited 8 years ago)
Original post by Indeterminate
Well here's my method:

Spoiler



Right. Yes, that works due to the z3z^3 that I'd forgotten about. That's a lot easier.

I have it in my mind that the "compute a series" approach is usually a lot more hassle than just using the formulae, but this is a nice counterexample.
Problem 548**

By considering a suitable integral, or otherwise, prove that

ln2<π212<1 \ln 2 < \frac{\pi^2}{12} < 1 .
(edited 8 years ago)
Original post by 16Characters....
Problem 548**

By considering a suitable integral, or otherwise, prove that

ln2<π212<1 \ln 2 < \frac{\pi^2}{12} < 1 .


A proof using series, because integrals are boring. :colone:

Spoiler

First note that:
n=11n2=π26\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{n^2} = \dfrac{\pi^2}{6}.
A reasonably well known fact, which can be checked using Parseval on f(x)=xf(x)=x.
(OK, that uses an integral, but there's a series too so shush yo. :tongue:)
Thus:
π212=n=112n2\dfrac{\pi^2}{12}= \displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n^2}

Also:
ln(1+x)=n=1(1)n+1xnn\ln(1+x)=\displaystyle\sum_{n=1}^{ \infty} \dfrac{(-1)^{n+1}x^n}{n}
converges for x<1|x|<1 and for x=1x=1. So:
ln(2)=n=1(1)n+1n=n=112n1n=112n=n=112n(2n1)\ln(2)=\displaystyle\sum_{n=1}^{ \infty} \dfrac{(-1)^{n+1}}{n}=\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n-1}-\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n}=\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n(2n-1)}.

But nN, 12n(2n1)12n2\forall n \in \mathbb{N}, \ \dfrac{1}{2n(2n-1)} \leq \dfrac{1}{2n^2} with equality if, and only if, n=1n=1.
Hence:
n=112n(2n1)<n=112n2[br]ln(2)<π212\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n(2n-1)}<\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n^2}[br]\Rightarrow \ln(2)<\dfrac{\pi^2}{12}

Now observe that: n=112n=1\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2^n}=1.
Note further that: nN, 12n12n2\forall n \in \mathbb{N}, \ \dfrac{1}{2^n} \leq \dfrac{1}{2n^2} with equality if, and only if, n=1n=1.
To prove this: Note the result is true for n=1,2n=1,2, and that if true for n=k2n=k \geq 2, then:
12k+1=12(2k)<12(k+1)2<1(k+2)2\dfrac{1}{2^{k+1}}=\dfrac{1}{2(2^k)}<\dfrac{1}{2(k+1)^2}<\dfrac{1}{(k+2)^2}.
The last inequality holds as:
k22>02k2+4k+2>k2+4k+42(k+1)2>(k+2)2k^2-2>0 \Rightarrow 2k^2+4k+2>k^2+4k+4 \Rightarrow 2(k+1)^2>(k+2)^2.

Thus we have that:
n=112n2<n=112n[br]π212<1\displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2n^2}< \displaystyle\sum_{n=1}^{ \infty} \dfrac{1}{2^n}[br]\Rightarrow \dfrac{\pi^2}{12}<1

So: ln2<π212<1\ln 2 < \dfrac{\pi^2}{12} < 1.
(edited 8 years ago)
Problem 549***

Suppose that MM is a non-negative matrix with dominant eigenvalue μ\mu. Let s>μs>\mu. Using the matrix expansion

(IMs)1=k=0(Ms)k\displaystyle\left(I-\frac{M}s\right)^{-1} = \sum_{k=0}^\infty \left(\frac{M}s\right)^{k}


Show that if u(0)1=1\left\Vert\mathbf{u}(0)\right \Vert _1 = 1 (the 1-norm of u(0)\mathbf{u}(0) is 1) then

k=0mksks(sIM)1u(0)1k=0pksk\displaystyle\sum_{k=0}^\infty \frac{m_k}{s^k} \le s\left\Vert (sI-M)^{-1}\mathbf{u}(0) \right \Vert _1\le \sum_{k=0}^\infty \frac{p_k}{s^k}


where mkm_k and pkp_k are the minimum and maximum column sums of MkM^k respectively.
Original post by joostan
A proof using series, because integrals are boring. .

Spoiler

Problem 550

Show that

0ln3xx2+1 dx=0\displaystyle \int_{0}^{\infty} \dfrac{\ln^3 x}{x^2 + 1} \ dx = 0
Original post by Indeterminate
Problem 550

Show that

0ln3xx2+1 dx=0\displaystyle \int_{0}^{\infty} \dfrac{\ln^3 x}{x^2 + 1} \ dx = 0


Solution:
ImageUploadedByStudent Room1448912864.669981.jpg


Posted from TSR Mobile
Original post by physicsmaths


That's the solution I was looking for :biggrin:

The complex analysis route is really quite messy in this case.

Quick Reply

Latest