an oddly satisfying sum
In this video I calculate the sum of sin((2n+1)x)/2n+1 which will lead us to a beautiful adventure involving power series, log, complex numbers, and integrals. Enjoy the ride!
Integral 1/1-x^2: • Integral 1/1-x^2 using...
KZread channel: / drpeyam
TikTok: / drpeyam
Instagram: / peyamstagram
Twitter: / drpeyam
Teespring merch: teespring.com/stores/dr-peyam
Пікірлер: 100
"i don't like to be in the bottom, but i don't care" cracked me up
@RamAurelius
2 жыл бұрын
*Bprp wants to know your location*
@dougr.2398
2 жыл бұрын
@@RamAurelius 😳😳😳😱😱😱😱😂🤣😂
The result π/4 is valid on (0, π). On (-π, 0) the value is instead -π/4. At x=kπ the result is 0. The sum is 2π-periodic in x.
@jelmerterburg3588
2 жыл бұрын
Yeah, you'd almost think that this infinite sum is a square wave ;)
At 5:11 we have to be more careful, when splitting up the series into two we can no longer guarantee convergence (as a simple counterexample, neither of them converge when x = 0, which is when the initial series evaluates to 0 and not pi/4).
Here's how I did it. Consider the exponential series sum from n=0 to infty of e^(i(2n+1)x)/(2n+1). The sine series we're interested in is the imaginary part. If we differentiate the exponential series term by term wrt x we get a geometric series which turns out to be real-valued. So the sine series has derivative 0 and hence is constant. Hence we can set x to a convenient value, say pi/2, and get 1-1/3+1/5-... = pi/4
@wolfbirk8295
Жыл бұрын
What about convergence ?....
Ah yes, the Fu-Yei Series. Another member of the family.
@bprpfast
2 жыл бұрын
Omg 😂
Highly satifying.
This is such a beautiful sum!
Applied math hat on killed me🤣
Thank you!
I really liked your videos and now with these emojis and all these jokes i love them even more hahaha. Great job man :D
Thanks for the video very interesting sum!
@silentintegrals9104
2 жыл бұрын
toally agree!!
This result needs a bit more care with the complex arguments. I remember experimenting with similar series as an undergrad and if i recall correctly these kind of functions are only piecewise constant with jumps at certain angles. In particular if you take the derivative you‘d get a sum of cosines which can be zero infinity and neg. infinity with the latter two cases at the positions of the jumps
Infinite series and complex numbers. What a cool pair!! :D
I'm amazed Dr. Peyam how you keep able coming up with such varied examples. Also the gong at the beginning made me chuckle a bit.
@drpeyam
2 жыл бұрын
Thank you!!
@dougr.2398
2 жыл бұрын
Going, GOING…. GONG!!! (= going minus « i »)
Thanks for watching
If you integrate with repect to x repeatedly and plug in nice values of x, you can get 1+1/2^k+1/3^k+... for k even and 1-1/2^k+1/3^k-... for k odd. 😁
@drpeyam
2 жыл бұрын
Beautiful!!!!
Are you trying to convince us that sin(0) + sin(3*0)/3 + sin(5*0)/5 + ... equals pi/4?
@drpeyam
2 жыл бұрын
Yep, on (0,pi)
Great result! But a little worry about those steps when substituting those complex entries, it needs to justify those argument angles 😶
Taking undergrad PDEs rn, recognized this sum from the thumbnail given how many times I’ve had to write down the Fourier series for 1 🤣
Capo total, mon ami! 😄😄😄
This sum is the Fourier series of the function f(x) = {-π/4 for -π < x < 0, +π/4 for 0 < x < +π}. So the result should be 0 for x = nπ, -π/4 for (2n-1)π < x < 2nπ, π/4 for 2nπ < x < (2n+1)π for any n integer.
@drpeyam
2 жыл бұрын
Yep, so it’s pi/4 on (0,pi), just as expected
3:22 HE SAID IT!!
Dear Dr Peyam. Decompose the function f(x)=π/4, given in the interval (0,π), into a Fourier series in terms of sines. You will get, π/4= ∑(from n= 1 to ∞)sin((2n-1)x)/(2n-1) =sinx+sin3x/3x +sin5x/5x...... On this path you will find an infinite number of "discoveries".
@drpeyam
2 жыл бұрын
Correct
Actually, you should use alternately the principal logarithm and the other one. I recognized it as the Fourier transform of a square wave.
awesome video
I first encountered this type of series in a boundary value problem
6:42 finally a moment in this video where I can relate with you😅🥰
Amazing inverse-Fourier deduction.
An oddly satisfying video
Perfect!! What about heavenly even satisfying sum?
simply amazing. this is the man that can find the last digit of π
Ja,ja,ja. Excelente! :D
Thanks
@drpeyam
2 жыл бұрын
Thanks so much for the super thanks!!! 😍😍😍
11:16 there should be e^(ix). I hoped the answer would've been complex :D
What if we considered a different branch cut at the end?
@drpeyam
2 жыл бұрын
Probably cancels out
As soon as you mentioned the power series for arctanx, I though of De Moivre's Theorem, (cosx+isinx)^n = cos(nx)+isin(nx) because then you could input e^(ix) into the arctan expansion and then write e^((2n+1)ix) as cos((2n+1)x) + isin((2n+1)x). The problem is that arctan alternates, so you probably would have to use artanh instead, so that might be more difficult. I haven't finished the video yet because my phone's at 2% but I'll be back to see if my prediction was useful or not.
@drpeyam
2 жыл бұрын
Nah, arctanh sucks lol
@violintegral
2 жыл бұрын
@@drpeyam well you really did use artanhx when you used the identity 1/iarctan(ix)=artanhx=1/2ln((1+x)/(1-x))
@wolfbirk8295
Жыл бұрын
@@violintegral how is arctan(z) defined ?...
I am from India and preparing for competitive exam. Your videos are so helpful to me ❤️❤️
Here's a challenge!! For alpha>1, study the convergence of the series sum(from n=3 to inf) [ int(from n to n+1) 1/(x^alpha * ln x) dx ] A series of an integral!!
@drpeyam
Жыл бұрын
Also a homework problem 😂
@ianmi4i727
Жыл бұрын
@@drpeyam A more or less difficult one!!
Lovely Math!
@drpeyam
2 жыл бұрын
Agreeeeed
@silentintegrals9104
2 жыл бұрын
Totally agree!!!
You look great Peyam 😊
@drpeyam
2 жыл бұрын
Thanks so much 🥰
Es muy genial
sorry but can I ask how many answers does ln(-1) has?
@drpeyam
2 жыл бұрын
Infinitely many, but one principal one
What’s the application to Physics ? (My perpetual question)
@drpeyam
2 жыл бұрын
Quantum mechanics lol
@drpeyam
2 жыл бұрын
Also this is the Fourier series of a step function
@dougr.2398
2 жыл бұрын
@@drpeyam please say a few words more? Thank you!
@dougr.2398
2 жыл бұрын
@@drpeyam what kind of a step function? A periodic one or thé Heaviside step function? (Or is that immaterial, somehow?)…..one that goes from zero to one over a finite interval or infinite (infinite interval might require Fourier intégrâmes, no?). Or one That goes from minus one-half to one-half, or minus one to plus one? (Some results then would/should differ by a constant then….)
eugh... even when i know its in complex numbers, writing "ln(-1)" makes me feel filthy
“i” don’t care if it’s top or bottom 😎
I is in denominator but I do not care. Hilarious joke, just love it. I think with knowing of inverse hyperbolic functions all of it would be faster ;) But this explanation just contains them implicitly so it's okay ;)
the results is incorrect for x=0, where did we use that x should be nonzero?
@drpeyam
2 жыл бұрын
x is in (0,pi)
@Czeckie
2 жыл бұрын
@@drpeyam sure, but why? I think the devil is in the detail at 8:15 as the logarithm has branch points at x=+-1.
@goblin5003
2 жыл бұрын
@@Czeckie 4:32 imagine what happens if x=0
No my love! if your sum started at n=0, you would get pi/4. Since it starts at n=1 you get pi/4 -sin(x)
How come? This sum depends on x. The result also has to depend on x
@drpeyam
2 жыл бұрын
Isn’t that amazing? It doesn’t depend on x, the function converges to a constant
@Unchained_Alice
2 жыл бұрын
@@drpeyam It seems to converge to -π/4 for some x values and to 0 for certain ones. The 0 one is clear but the negative one I just tested on my computer so I can't be sure. Is the negative one due to the principal logarithm not being continuous for negative reals?
@drpeyam
2 жыл бұрын
Right, it’s still pi/4 on (0,pi) so it’s ok
you must have had fun editing this video
@drpeyam
2 жыл бұрын
I really did hahaha 😝
@drpeyam
2 жыл бұрын
I loved the cow sound
What was that :0
Yes!, our little friends i's :)
Instead of step 3 I would use arctan a - arctan b = arctan ((a-b)/(1+ab))
@drpeyam
2 жыл бұрын
I heard that might make the denominator 0 though
@AlexAlex-kb4js
2 жыл бұрын
@@drpeyam sure, but arctan (something/0) = arctan (infinity) = pi/2 And we have some 1/2 in front so it should lead as well to pi/4
Yoooooooooooooooo!
Not right demonstration. ln(-1)=ln(exp(i*Pi))=ln(exp(3i*Pi)=ln(exp((2k+1)i*Pi)). But that doesnt mean that i*Pi=3i*Pi..
@drpeyam
2 жыл бұрын
The demonstration is correct, principal logs
Proffesor,,,you will not be able to use the arctangent formula because....here xy =-1 ,,,but tan-1(x) + tan-1(y) = tan-1((x-y)/(1+xy))
5:14 .... .. . . . ... . . . . . . .. . .. . . Wtf
@drpeyam
2 жыл бұрын
I was hoping someone would notice 😂😂😂😂😂
Hola
Those edits 😂
Should a 7th grader be here?
@blableu4519
2 жыл бұрын
why not