∫ 0 ∞ ( x sin x ) 5 d x
If the integral above is equal to b a π for coprime positive integers a and b , find the value of a + b + 1 .
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.
Good observation about the integration by parts. You should be clearer about what u , v functions you are using, to explain what "works here (not everywhere)" means.
When evaluating similar equations, it's typically better to find a general method, instead of methods that work for individual equations. As an example, it is better to use sin 5 x = 1 6 1 ( sin 5 x − 5 sin 3 x + 1 0 sin x ) , which gives us an easy way to evaluate ∫ x sin 5 x without knowing what ∫ x sin 3 x is.
No need of "boring" IBP, I believe. IBP becomes very "boring" as you go forward. All your lemmas can be proved using Feynman's way as well.
Log in to reply
I don't know that. You can do it using contours too!
Log in to reply
Yeah, I know. I am going to post a solution using that or some other better way in near future.
PS: Feynman's way is just "Differentiation under integral sign"
@Calvin Lin Sir, thank you for the feedback. Next time I'll make sure I use something similar...
First, for general indefinite integral
∫ x 5 S i n ( a x ) d x
We use the recursive relations
∫ x n S i n ( a x ) d x = − ( n − 1 ) x n − 1 S i n ( a x ) + ( n − 1 ) a ∫ x n − 1 C o s ( a x ) d x
∫ x n C o s ( a x ) d x = − ( n − 1 ) x n − 1 C o s ( a x ) − ( n − 1 ) a ∫ x n − 1 S i n ( a x ) d x
to derive the following relation
∫ x 5 S i n ( a x ) d x = 2 4 x 4 1 ( a 3 x 3 C o s ( a x ) + b 2 x 2 S i n ( a x ) − 2 b x C o s ( a x ) − 6 S i n ( a x ) ) + 2 4 b 4 ∫ x S i n ( a x ) d x
Now, using trigonometric identities, we have the following relation
( S i n ( x ) ) 5 = 1 6 1 ( 1 0 S i n ( x ) − 5 S i n ( 3 x ) + S i n ( 5 x ) )
so that, using both relations, we have
∫ 0 ∞ ( x S i n ( x ) ) 5 d x = L i m i t ∣ f ( x ) , x → ∞ ∣ − L i m i t ∣ f ( x ) , x → 0 ∣ + 1 9 2 1 1 5 ∫ 0 ∞ x S i n ( x ) d x
where f ( x ) is a sum of terms of the form
c x n b S i n ( a x ) o r c x n b C o s ( a x )
where 0 < n < 5 , and a , b , c are integers. Using L’Hospital’s rule, all of the terms vanish at either limit, i.e., we have f ( 0 ) = 0 and f ( ∞ ) = 0 , and so we are left with
∫ 0 ∞ ( x S i n ( x ) ) 5 d x = 1 9 2 1 1 5 ∫ 0 ∞ x S i n ( x ) d x = 3 8 4 1 1 5 π
Note that with this approach, which can be generalized for other n , we really only need to pay attention to the coefficients of the Dirichlet Integral function and ignore the rest, keeping in mind the following detail
∫ 0 ∞ x S i n ( a x ) d x = ∫ 0 ∞ x S i n ( x ) d x = 2 π
.
Edit
: Using the following trigonometric identity, for odd
n
, for those coefficients
( S i n ( x ) ) n = 2 n 2 k = 0 ∑ 2 n − 1 ( ( − 1 ) ( 2 n − 1 − k ) k ! ( n − k ) ! n ! S i n ( ( n − 2 k ) x ) )
and using the approach outlined above for n = 5 , this works out to, for odd n
∫ 0 ∞ ( x S i n ( x ) ) n = 2 n ( n − 1 ) ! π k = 0 ∑ 2 n − 1 ( ( − 1 ) k k ! ( n − k ) ! n ! ( n − 2 k ) n − 1 )
Yes, this does look like Sharma's result, except for the coefficient on the left side of the ∑ . The coefficient here is sufficient, no need to employ a power of i and the S i n e function. And, of course , ( n − 1 ) ! = Γ ( n )
Great! Thanks for the generalization as well.
I have the same idea of IBP's but is slightly long winded.
How do you prove/derive the trigo identity:
( S i n ( x ) ) n = 2 n 2 k = 0 ∑ 2 n − 1 ( ( − 1 ) ( 2 n − 1 − k ) k ! ( n − k ) ! n ! S i n ( ( n − 2 k ) x ) ) ?
Log in to reply
You can make use of trigonometric functions in exponential form to arrive at this identity. How this can be done should be the subject of a note or a wiki, since the approach is very general, not only for powers, but for multiples, i.e. ( S i n ( x ) ) a and S i n ( a x ) , for any trigonometric function. I used this method to derive the general expansion of T a n ( a x ) in here Updating Archimedes
Also, didn't Calvin Lin give a lecture about Chebyshev coefficients , which is a related subject?
For this identity here that we're talking about, I simply looked up a table of trigonometric identities.
Meanwhile, Sharma should fix the coefficient on the left side of the ∑
Log in to reply
Also, didn't Calvin Lin give a lecture about Chebyshev coefficients, which is a related subject?
I don't think I was there. I'm not a moderator that time.
It's still surprising that you've condensed my 3 page long solution into a (about) 1 page solution. Huge appreciation!
Surely better than mine! Nicely explained!
Log in to reply
Kartik, you might be missing the point that your solution suggests how to derive the equation for powers of the Sine function. That's what I find interesting about it.
I will try to find an expression for the general integral.
∫ 0 ∞ ( x sin x ) n d x
First let's consider that n is odd. Then,
∫ 0 ∞ ( 2 i ) n x n ( e i x − e − i x ) n d x
− ( 2 i ) n − 1 ∫ 0 ∞ 2 i 1 ( 1 − e 2 i x ) n e i n x x n d x
− ( 2 i ) n − 1 ∫ 0 ∞ 2 i 1 k = 0 ∑ n ( k n ) ( − 1 ) k e ( 2 k − n ) i x x n d x
As the first case is for odd n ,
− ( 2 i ) n − 1 ∫ 0 ∞ 2 i 1 k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k ( e ( 2 k − n ) i x − e − ( 2 k − n ) i x ) x n d x
− ( 2 i ) n − 1 ∫ 0 ∞ k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k sin ( ( n − 2 k ) x ) x n d x
Substitute x = u
− 2 n i n − 1 ∫ 0 ∞ k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k u sin ( ( n − 2 k ) u ) u 1 − n / 2 − 1 d u
Now, we can use Mellin transform and hence Ramanujan Master Theorem.
Here, for the generating function, we'll use x sin ( x ) = i = 0 ∑ ∞ ( 2 i + 1 ) ! ( − u ) i . And our generating function would become
α ( u ) = k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k m = 0 ∑ ∞ Γ ( 2 m + 2 ) ( n − 2 k ) 2 m + 1 Γ ( m + 1 ) m ! ( − u ) m
= m = 0 ∑ ∞ ⎝ ⎛ Γ ( 2 m + 2 ) Γ ( m + 1 ) k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k ( n − 2 k ) 2 m + 1 ⎠ ⎞ m ! ( − u ) m
so that ϕ ( m ) = Γ ( 2 m + 2 ) Γ ( m + 1 ) k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k ( n − 2 k ) 2 m + 1
Now, using RMT, our integral becomes
= − 2 n i n − 1 Γ ( 2 2 − n ) ϕ ( 2 n − 2 )
= − 2 n i n − 1 Γ ( 2 2 − n ) Γ ( n − 2 + 2 ) Γ ( 2 n − 2 + 1 ) k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k ( n − 2 k ) n − 2 + 1
Finally,
∫ 0 ∞ ( x sin x ) n d x = − 2 n i n − 1 sin ( 2 π ( 2 − n ) ) Γ ( n ) π k = 0 ∑ ( n − 1 ) / 2 ( k n ) ( − 1 ) k ( n − 2 k ) n − 1 , for odd n
Exercise for the reader : Try to extend it for even n [Hint: Only one more(or less) term will get added to the binomial expansion and rest the same].
Anyways, I will be posting it but it is good to exercise while reading.
Wonderful solution.
Thank you. Let me print this solution!
@Pi Han Goh What do you think?
I never wanted to do this, honestly. I was thinking of Contour for days but I failed as I never got the full solution. So, I had to get back to Mellin. BTW, can you do anything as the solutions page of this problem has become very long?
Log in to reply
I think this is great. I can't verify that it's 100% correct because I haven't mastered RMT.
I don't think it's that long. Considering that you have generalized this for all odd n , I think it's much better than any of the solutions provided, because they have only provided the solution for n = 5 (and for n = 3 by Kishore).
Log in to reply
@Pi Han Goh :P, I don't expect to figure this right now...
No, I am not saying that my solution is long. I am saying that this problem has got many solutions(though, only 4) but more significantly, many comments on the solutions so that it has become quite difficult to scroll down. So, if you can disable Lu Chee Ket's solution(if he desires or if the discussion is over) or whatever you think might be correct.
Yeah, I also can't verify if it's 100% correct. It seems to verify for the values but I don't know.
BTW, can it have a better form? And, are you trying for even n ?
Log in to reply
@Kartik Sharma – Let me delete his solution and his comments.
You might be interested in equations 37 and 38 .
I haven't completely understood your solution yet so it's unlikely that I can solve for even n .
Log in to reply
@Pi Han Goh – Thanks a lot for linking me to that! Although the solutions to the results are almost obvious from my solution, but I really appreciate the form they are given in. Great!
Eulerian number, hmm. Well, can you make a problem using Eulerian number and stuff? I would love to see how they are used.
@Pi Han Goh – We can have some idea onto almost any definite integral almost immediately with concept of general numerical values to them in mind. Coincidence of solvable or not is not compulsory.
@Kartik Sharma – You shall find trouble to evaluate (Pi/ 2)/ 999! Sum [(-1)^i (1000 i)(500 - i)^999] for i = 0 to 500, for n = 1000, without calculator of many significant figures. But 35 Pi/ 1602 is an almost immediate's estimation with calculator of only 15 significant figures.
Hey, in RMT, can we take ϕ ( k ) to be any function of k ?
Log in to reply
Yeah, it can be any function. But what will matter is ϕ ( − s ) , so keep the limit of ϕ ( k ) at s intact.
sinc x is actually meant sin (x Pi)/ (x Pi) from the book I read, Kishore S Shenoy.
For n = 1000, 35 Pi/ 1602 is an immediate estimation. So the meaning of this question is just to study the fundamental thing since they are all solved. Numerical method is still a general method for example when we obtain a value of integral = - Euler's number of - 0.5772+.
Problem Loading...
Note Loading...
Set Loading...
First, some introduction...
We know sin 3 x = 4 3 sin x − sin 3 x ∫ 0 ∞ x sin 3 x d x = 4 1 [ ∫ 0 ∞ x 3 sin x d x − ∫ 0 ∞ x sin 3 x d x ] = 8 3 π − 4 1 ∫ 0 ∞ y sin y d y = 4 π
We know sin 5 x = 1 6 sin 5 x + 2 0 sin 3 x − 5 sin x ∫ 0 ∞ x sin 5 x d x = 1 6 1 [ ∫ 0 ∞ x sin 5 x d x + 2 0 ∫ 0 ∞ x sin 3 x d x − 5 ∫ 0 ∞ x sin x d x ] = 1 6 1 [ 2 π + 5 π − 2 5 π ] = 1 6 3 π
So, we can use the following here(not everywhere). You can use this here because x → ∞ lim ∫ v ( x ) d x = x → ∞ lim x n 1 = 0 , for some n > 0 and u ( 0 ) = sin 0 = 0 So, ∫ 0 ∞ u ( x ) v ( x ) d x = − ∫ 0 ∞ ( u ′ ( x ) ∫ v ( x ) d x ) d x
∫ 0 ∞ x 3 sin 3 x d x = 2 3 ∫ 0 ∞ x 2 sin 2 x cos x d x = 2 3 ∫ 0 ∞ x 2 sin x cos 2 x − s i n 3 x d x = 2 3 [ 2 ∫ 0 ∞ x sin x d x − 3 ∫ 0 ∞ x sin 3 x d x ] = 2 3 ( π − 4 3 π ) = 8 3 π
Let's do Parts! ∫ 0 ∞ ( x sin x ) 5 d x = 4 5 ∫ 0 ∞ x 4 sin 4 x cos x d x = 1 2 5 ∫ 0 ∞ x 3 4 sin 3 x cos 2 x − sin 5 x d x Using cos 2 x = 1 − sin 2 x ∫ 0 ∞ ( x sin x ) 5 d x = 1 2 5 [ ∫ 0 ∞ x 3 4 sin 3 x d x − ∫ 0 ∞ x 3 5 sin 5 d x ] Using ∫ 0 ∞ x 3 s i n 3 x d x = 8 3 π , ∫ 0 ∞ ( x sin x ) 5 d x = 8 5 π − 1 2 2 5 ∫ 0 ∞ x 3 sin 5 x d x Repeating the Integration By Parts again, ∫ 0 ∞ ( x sin x ) 5 d x = 8 5 π − 2 4 1 2 5 [ ∫ 0 ∞ x 4 sin 3 x d x − ∫ 0 ∞ x 5 sin 5 x d x ]
Using ∫ 0 ∞ x sin 3 x d x = 4 π and ∫ 0 ∞ x sin 5 x d x = 1 6 3 π
So, ∫ 0 ∞ x 4 sin 3 x d x − ∫ 0 ∞ x 5 sin 5 x d x = π − 1 6 1 5 π = 1 6 π
⇒ ∫ 0 ∞ ( x sin x ) 5 d x = 8 5 π − 2 4 × 1 6 1 2 5 π = 3 8 4 1 1 5 π
∴ ∫ 0 ∞ ⎝ ⎜ ⎜ ⎜ ⎛ x sin x ⎠ ⎟ ⎟ ⎟ ⎞ 5 d x = 3 8 4 1 1 5 π
So, a + b + 1 = 1 1 5 + 3 8 4 + 1 = 5 0 0