f ( x ) = k = 1 ∑ ∞ k 2 sin ( k x )
Find the value of x , with 0 < x ≤ 2 π , where f ( x ) attains its maximal value. Write x = b a π , where a and b are coprime positive integers, and enter a + b .
If you come to the conclusion that no such x exists, enter 666.
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.
@Otto Bretscher Fun question. I really should expand on some of the details here but my solution is long enough already and I've lost my enthusiasm at this late hour. :)
Log in to reply
Thank you for another clear and detailed solution, Brian! (+1) This is exactly what I had in mind; the key observation is f ′ ( x ) = − ln ( 2 sin ( 2 x ) ) for 0 < x < 2 π .
The function we study here is known as a Clausen function , more precisely, it is
S 2 ( x ) = C l 2 ( x ) = k = 1 ∑ ∞ k 2 sin ( k x ) = − ∫ 0 x ln ( 2 sin ( t / 2 ) ) d t
for 0 ≤ x ≤ 2 π . (The last expression is Clausen's Integral .)
The maximum of S 2 ( x ) is attained at x = 3 π , where the integrand changes from negative to positive.
I thought this would be more simple, here was my solution: if x = .000000001...~6.28 then there is no possible x, x has to be over 13 to be existent in this equation, but after seeing Brian's work it came clear to me I have a bit more reaserch to do because I'm 13 and technically on my own learning this stuff :/
Log in to reply
I'm impressed that you are teaching yourself this kind of material at the age of 13. You are way ahead of where I was at when I was your age. :)
Log in to reply
my inspiration is jacob barnett, hes a 16 year old autistic child genius with a iq higher than einstein, i am autistic as well and i want to use my brain to be the best that i can at math, logic, and computer science as soon as i can. i have alot of reaserch to do though ;)
Log in to reply
@Garrett O’Brien – I admire your desire to make the best use of your mind. I hadn't heard of Jacob Barnett before so I watched a TEDxTeen talk by him. He's a terrific role model for you. :)
how to prove k = 1 ∑ ∞ k z k = − ln ( 1 − z ) . ?
Log in to reply
This is the Maclaurin series for the natural log function, extended to complex numbers for ∣ z ∣ < 1 .
For a more direct proof, start with 1 − t 1 = 1 + t + t 2 + t 3 + . . . .
Then − ln ( 1 − z ) = ∫ 0 z 1 − t 1 d t = ∫ 0 z ( 1 + t + t 2 + t 3 + . . . ) d t
⟹ − ln ( 1 − z ) = z + 2 z 2 + 3 z 3 + 4 z 4 + . . . = k = 1 ∑ ∞ k z k .
Log in to reply
So it is valid for complex number also ?
Log in to reply
@Kushal Bose – Yes, with the domain of convergence being a disc on the complex plane of radius 1 . A relevant discussion is given in section 1.5 of this paper .
Good solution, but you made a straight forward problem look complicated.
Log in to reply
Brian is skillfully outlining a proof of the fact that Clausen's integral represents the Clausen function S 2 ( x ) , which is not straightforward.
Problem Loading...
Note Loading...
Set Loading...
By the Weierstrass M-test f ( x ) converges uniformly for any x , and since f ( x ) is also the sum of differentiable functions f ′ ( x ) exists with f ′ ( x ) = k = 1 ∑ ∞ k cos ( k x ) . By the Dirichlet test for convergence we can conclude that f ′ ( x ) converges for 0 < x < 2 π .
Now cos ( y ) = 2 e i y + e − i y , so
f ′ ( x ) = k = 1 ∑ ∞ k cos ( k x ) = k = 1 ∑ ∞ 2 e i k x + e − i k x = 2 − ln ( 1 − e i k x ) − ln ( 1 − e − i k x )
since in general k = 1 ∑ ∞ k z k = − ln ( 1 − z ) .
(There are radius of convergence issues here to deal with, but for now I will just carry on .....)
So f ′ ( x ) = − 2 1 ln ( ( 1 − e i k x ) ( 1 − e − i k x ) ) = 2 1 ln ( 2 − 2 cos ( x ) ) = − ln ( 2 sin ( 2 x ) )
after making use of the identity cos ( x ) = 1 − 2 sin 2 ( 2 x ) . Now f ′ ( x ) = 0 when
2 sin ( 2 x ) = 1 ⟹ sin ( 2 x ) = 2 1 ⟹ 2 x = 6 π + n ∗ 2 π or 6 5 π + n ∗ 2 π
⟹ x = 3 π + n ∗ 4 π or x = 3 5 π + n ∗ 4 π .
So in the given interval the extrema occur at x = 3 π and x = 3 5 π . Plugging these values back into f ( x ) we see that the first two terms, (which dominate the sum as the denominators are the least in the series), with x = 3 π are positive and with x = 3 5 π are negative, so (informally) we can anticipate that the former critical point represents a maximum and the latter a minimum. Thus the maximal value of f ( x ) on the given interval occurs at x = 3 π , giving us a + b = 1 + 3 = 4 .