so you want a HARD integral from the Berkeley Math Tournament
You can now register for BMT 2022 here: www.ocf.berkeley.edu/~bmt/bmt...
The first-place team wins $1000 bprp scholarship and I will be there!
We will integrate x/tan(x) from 0 to pi/2. We will use Feynman's technique (aka Feynman's trick of integration or differentiation under the integral sign) to compute this integral. This problem is from the Berkeley Math Tournament in 2020. Here's the link to the exam and the solution: www.ocf.berkeley.edu/~bmt/arc...
If you are also interested in sponsoring BMT and making an impact, please contact communications@bmt.berkeley.edu
Thank you!
0:00 we haven't done a hard integral for a while
2:33 the steps of Feynman's trick of integral
5:30 differentiate I(a) first, then do the integral
18:59 integrate I'(a) with respect to a to get I(a)
🛍 Shop math t-shirts & hoodies: bit.ly/bprpmerch
10% off with the code "WELCOME10"
----------------------------------------
**Thanks to ALL my lovely patrons for supporting my channel and believing in what I do**
AP-IP Ben Delo Marcelo Silva Ehud Ezra 3blue1brown Joseph DeStefano
Mark Mann Philippe Zivan Sussholz AlkanKondo89 Adam Quentin Colley
Gary Tugan Stephen Stofka Alex Dodge Gary Huntress Alison Hansel
Delton Ding Klemens Christopher Ursich buda Vincent Poirier Toma Kolev
Tibees Bob Maxell A.B.C Cristian Navarro Jan Bormans Galios Theorist
Robert Sundling Stuart Wurtman Nick S William O'Corrigan Ron Jensen
Patapom Daniel Kahn Lea Denise James Steven Ridgway Jason Bucata
Mirko Schultz xeioex Jean-Manuel Izaret Jason Clement robert huff
Julian Moik Hiu Fung Lam Ronald Bryant Jan Řehák Robert Toltowicz
Angel Marchev, Jr. Antonio Luiz Brandao SquadriWilliam Laderer Natasha Caron Yevonnael Andrew Angel Marchev Sam Padilla ScienceBro Ryan Bingham
Papa Fassi Hoang Nguyen Arun Iyengar Michael Miller Sandun Panthangi
Skorj Olafsen Riley Faison Rolf Waefler Andrew Jack Ingham P Dwag Jason Kevin Davis Franco Tejero Klasseh Khornate Richard Payne Witek Mozga Brandon Smith Jan Lukas Kiermeyer Ralph Sato Kischel Nair Carsten Milkau Keith Kevelson Christoph Hipp Witness Forest Roberts Abd-alijaleel Laraki Anthony Bruent-Bessette Samuel Gronwold Tyler Bennett christopher careta Troy R Katy Lap C Niltiac, Stealer of Souls Jon Daivd R meh Tom Noa Overloop Jude Khine R3factor. Jasmine Soni L wan na Marcelo Silva
----------------------------------------
💪 If you would also like to support this channel and have your name in the video description, then you could become my patron here / blackpenredpen
You can now register for BMT 2022 here: www.ocf.berkeley.edu/~bmt/bmt-2022/ The first-place team wins a $1000 bprp scholarship and I will be there!
Please solve -{integral (0-->π/2) {log[sinx]}dx}
@@orangesite7625 cos c + tan 90
what language is this?
Thank you very much for such a valuable content. I have a question why did you so casually change the conditions of a from greater than 0 to greater or equal to 0. Shouldn't you have returned to the integration part to see if it doesn't change anything? Indeed it doesn't but still ...
@@orangesite7625 answer is in this video but it can be done easier
And then multiply everything by BMT2020
Equate to BMT2021 and solve for x 🤣
😂😂😂😂
🤣🤣🤣
😂
You clown 🤣🤣🤣🤣
Hello! You probably don’t remember me but I took one of your classes back in 2016. I was struggling with math and you advised me to refer to your KZhead channel. It’s crazy how much this channel has grown. Congrats!! 🎉
Thank you!! I do recognize this name. And wow, it’s been over 6 years!! Hope you are doing well 😃
@@blackpenredpen where do you teach?
@@sphrcl. harvard
Op
@@sphrcl. but how is your math now 💀💀
I have another way of doing it I = Integ (x cos x / sin x) dx = Integ x/sin x d (sin x) = Integ x d ln (sin x) Applying integration by parts, I = [pi/2 ln sin (pi/2) - 0 ln sin (0)] - Integ (ln sin x) dx The integral from 0 to pi/2 of ln sin x dx is a classic integral (can be calculated by king’s property)
I also used this integration by parts.For the integral from 0 to pi/2 of ln sin x dx I write sinx as 1/(2 i )*exp[i x]*(1-exp[-2 i x]) If you take the ln you get - ln 2- π i/2+i x+ ln[1-exp[-2 i x]. Next you use that ln[1- z] = sum (z^k/k ),k =1 to inf..Here z=exp[-2 i x]. If you now integrate term by term it is easy to see that you get 0 . The rest of the integration for the other terms is then trivial and you get the answer - π/2 ln2 .
I forgot a detail : because of the symmetry of sin x with respect to π/2 you can take 1/2 of the integral from 0 to π. This makes the last integral 0.
That's the way I was going to go, but didn't see the ln.
I did the same way
Yeah, that was my first reaction as well. Glad to see someone else had the same thought process.
Excelent! You should post more challenging ones like this
I actually solved this one on my own! I would never have been able to do that without guys like you teaching.
Op
by the same method?
@@user-ox2qt2cm4c yeah, same method. I knew where to apply it though!
Same, Ibdidnt use at the exact same place but i knew i would need this technique
Check out BMT official solution: www.ocf.berkeley.edu/~bmt/wp-content/uploads/2022/04/calculus-solutions-1.pdf
We can use integration by parts, x is the one to be differentiated and cotx must be integrated, after doing that we get I = -int(lnsinx) from 0 to pi/2 which is a standard integral and we get I = (pi/2)ln2
Hey in the future videos could you explain why cos(π/2^2)•cos(π/2^3)•...•cos(π/2^(n+1))=1/2^n•sin(π/2^(n+1))??? Because i saw it in the solutions of a problem and i can't explain myself why, and also why does its limit tends to 2/π?
@blackpernredpen Can you actually solve this ? Pr(m>=N/2) = sum from m=N/2 to N of (n/m)* 0.25^m * 0.75^(n-m)
if n>0 , a>0 و a=0 ---> integral from 0 to (pi/2)^1/n of arctan(a.tan(x^n))/tan(x^n) dx =pi/2.ln(a+1)........... Is this true?!
Just finished my first semester in Calculus 1, and now I finally understand a lot of the concepts presented in this video, as if it summarized my entire semester into one big challenge (the only thing I haven't really learned yet is the partial derivatives, but they seem to make sense from the video) This was an AMAZING experience!!
I really enjoyed that. Best part of my day so far :D
Another beautiful use of Feynman's technique!!
Gracias por existir este canal..... Es de lo mejor que he encontrado. ..
I just love it, when pi and e show up together. Well, actually it makes sense… The pi comes from the trig-function and the log_e(2) from the integration.
Watching this gives me so much relief🤩
wow wow , u explained it in a very simple way thank you ! please do more videos
I love this channel. I hated calculus and did really poorly in it. Found this channel and it reignited my love for maths. Now I can follow along and even do the more difficult integrals thanks to these videos. Thank you a lot for igniting that flame again!!
Can you tell me how to start from scratch i want to learn calculus but i am bad at it
@@Legend-mr6ci my advice would be 1) try to make it fun for yourself by seeing where the topics are actually used in the real world, 2) go very slowly and take your time learning and understanding the concepts. Watch as many videos as possible and 3) practice a lot. So find videos you can practice along with. Calculus is a toolbox and if you don’t polish your tools in the toolbox then they won’t work that well
If you treat the integral as xcotx Then apply the power series of expansion for cotx we have the integral as: x ( 1/x−x/3−x^3/45−2x^5/945 - ⋯) After multiplying the x into the expansion, we can then easily integrate and evaluate the function from 0 to pi/2 and see it converges to pi/2 ln2 (By factoring Pi/2 out and then consider power series of ln functions)
This was my first thought! But I enjoyed the solution that he presented quite a lot!
this is a much better solution and answer can be obtained easily even by using byparts
Wow, Maclaurin series are good
There is one serious downside that no one here is quite pointing out, but assuming that you are not provided this in the competition, then you are asking someone to be able to either recite the Maclaurin series for cot x or to be able to reproduce it on their own. Which is an incredibly impractical thing to do. You appear to need to know Bernoulli numbers etc. to reproduce this expansion, or just memorise the not so friendly fractions here. You can't empirically produce cot x as easily as you would for say sin x - unless of course you know of a more succinct and beautiful way of doing it then what I seen on in the Internet.
This was a very, very satisfying demonstration BPRP. I didn't know about the Feynman technique until recently. Thank you!😸
Op
Excellent problem and wonderful explanation!
A different way: integral = int of x cotx dx, then using By Parts, to get - int of ln sin x dx, by using the property of int of f(x) = int of f（pi/2 - x）, we have 2I = int of (ln sinx + ln cosx) dx. then follow some basic algebra and trig, we get I =( pi/2)(ln2) detail: kzhead.info/sun/idmuZN1wjIyOfWw/bejne.html
When he said “you give it a go,” I did this exact technique!
Exactly so.
Exactly I solved the same
I was looking for this comment Thanks
You explain so well that i want all day watch you)
I've done this integral with a little bit different technique. First of all I've done IBP in first place to get "- int from 0 to pi/2 of ln(sinx) dx". Of course this integral also not so easy, and I've done separation of integration ranges from 0 to pi/4 and from pi/4 to pi/2. On second integral it was a u-sub like "x = pi/2 - u", to get from sine function - cosine. There is integral of sum of logarithms of sine and cosine functions from 0 to pi/4. Using log identities, we can get "int from 0 to pi/4 of ln(sin(2x)/2) dx", and with a u-sub "u = 2x". Assuming algebraic-integral equation "J = int from 0 to pi/2 ln(sin(x)) dx" and "J = -pi/2 ln2 - 1/2 J", we can get the same result as in this video. I've tried this first :D
This was also my method, but since int ln(sin(x)) dx = int ln(cos(x)) dx (for this region, using u=pi/2-x), I let 2A = ln(2*sin(x)*cos(x)) - ln(2) dx :)
@@yoto60 I can't understand🤔
how did you apply IBP on this x/tanx dx mate? Did you convert it to ∫x tan-1x dx
@@wondersoul9170 ooh thanks broo😁🙏❤️
@@shiviarora4173 of course not, its just x * cotx, x to be differentiate, cotx to be integrate
22:20 we can see the passion and appreciation this gentleman has for maths with just a look, he found his happiness and that's bittersweet af, nice video as always fella.
thought of applying beta fn. but who knew. hands down! i was literally smiling at the end 🙌
Nicely done 👍💯
It's always good to see a nice calculus problem......
Really interesting. Well done!
Beautiful integral
This was very cool to watch!
Very intersting solution -using a nice trig manipulation and the usual Differentiation Under Integral Sign (aka the "Feynman Trick"). But what I liked best here was the way you decomposed the irreducible quadratic factors (1+a^2u^)*(1+u^2) into linear fractions- by cleverly substituting the (a*u) term as a linear term -enabling to use the "cover-up" to get the constants "A" and "B" very easily. I watched your detailed videos on "Partial Fractions"- and the "cover up" short-cut method- never heard of it during my student days (I am a 60+ engineer). Enjoy your videos and it makes me feel young again. Thanks a ton BPRP, and also your partner Dr Peyam's videos-espeacially the integration ones.
It is questions like these that just make me very happy
Just wanted to say that I have subscribed to your channel. I've seen your videos since 2020, and thanks to you I learned a lot of cool math things I didn't know beforehand. =)
Thank you!
I would have used the substitution u = tan(x) first, leading to \int_0^{\infty} arctan(u)/(u (1+u²)) du. Then in that integral, replace the arctan(u) with arctan(a*u), so you have an integral depending on a. Differentiating with respect to a then gives the same integral as in the video at about 12:10. I think that's more intuitive than coming up with the trick x = arctan(tan(x)) first, and additionally, coming up with the idea that one should insert a factor of sec²(u).
You don't really need to multiply top and bottom by f sec²(u). Just substitute u=tan x straight off.
Thank you sir.
Very nice solution!
This type of integration is also known as integration by reduction formula here in India. Nice video btw!
I think reduction formula is something like I n = a I (n-1)+ b I (n-2) which gives a recursive relation among the integrals but this question has nothing to do with it
No it’s Leibniz rule but this one east with king property
You can use integral x* cotg x, then by parts, you have then - integral log( sinx), It's also on internet And some tric to use.
An alternative method is to integrate by parts with u = x and dv = cotxdx which will result in the integral of ln|sinx|dx over the same interval which can be easily evaluated using power series and setting y = 1 - sinx. The challenge is showing that the resultant integral converges to pi/2ln2, but honestly, I couldn't really care as long as you're able to show a solution.
Sin palabras... ¡genial!
More like this please
this was enlighting!
Thank you sir
Added it to my Math YT video collection!
I'am in high school I understand only the derivations and integrations but this is just.. wow. Thanks for such great content and the reassurance that I want to do math
Lol me watching this after not even being one semester through calc in high school and wanted to do a premed biochem major anyways: 😂
Differentiations*
@@yoelit3931 you are right mb, funny thing is I still don't understand, isn't feynman technique?
This reminded me of another hard one I can't forget, integral of (sqrt 1 + x^2)/x (non-trig solution takes up whole board) U already did that, but only trigonometrically, I'd like to see u do it with the "u world" as well, it's a bit harder that way, and I'd love it.
Well played
Hi, I have another way of doing it. we can change the 1/Tan(x) part to cot (x) and then Integrate X.Cot(x) using integration by parts. It would've left us with integral of ln|Sin(x)| from 0 to pi/2 which can be easily integrated using king's Rule further. :)
I was just telling someone yesterday about Feynman "re-discovering" differentiating under the integral, but I'd never seen it in action. This was awesome!
I came to add that this is just the “Feynman Rule” but you beat me to it :) When you’re stuck just try integrating under the integral. Worst case you get a different problem you still can’t solve :p
@@robertkb64 i need to rewatch this or another video explaining this concept again. it's been awhile haha
it appears to me that Feynmann found an undiscovered or rarely used way to use the Leibniz rule for differentiating under the integral. More generally, the Leibniz rule incorporates differentiation of the two limits (if they aren't constants).
Technically we can’t let a=0 at the end because then the argument from earlier doesn’t work so maybe we can take a limit as a approaches 0 to get the value for c
we can since if we take a to be greater than or equal to 0 from the start(as we should) when we get to the point in the limit, we don't calculate it we just say ''well if a=0 then we go back to the integral and see that I(0)=0 so for a>0.......''
@@paokaraforlife problem is if we just say a is greater than OR equal to 0 then when evaluating inverse tan of au as u approaches infinity we can’t do what we do in the case where a=0
@@Happy_Abe yeah and we don't We take the value and put it in the integral definition
@@paokaraforlife sorry I’m confused From what I understand the integral can’t be evaluated at a=0, that is not within the domain of the I(x) function as previously mentioned. We can’t just evaluate functions at points outside the permitted domain
@@Happy_Abe ok I'll take it from the beginning We define the function using he integral for a greater than or equal to 0 For a=0 we plug the value into the integral and we get the integral of 0 which is 0 For a>0 we do the same work as in the video and get a value through the limit
Powerful!
Kings property of integration and Integration by parts result: x*ln(sinx)|(0->π/2)-(0->π/2)definite integral ln(cosx)dx Where x*ln(sinx)|(0->π/2)=0 And -(0->π/2)definite integral ln(cosx)dx = -1/4*dβ/dn(1/2,1/2) = -1/4*β(1/2,1/2)*[ψ(1/2)-ψ(1)] = -1/4* Γ(1/2)^2/Γ(1)*[-2ln2-γ-(-γ)] = -1/4*π*[-2ln2] = πln2/2
excellent work
there was a pretty easy solution . we could use the product rule for integration by treating x as first function and cotx as second it would come out to be a-b where a = xlnsin(x ) evaluated from 0 to pi/2 ( it comes out to be zero) b= integral of lnsin(x) from 0 to pi/2 which is -pi/2*ln(2) so answer becomes -(- pi/2*ln(2))
Its basically an easy question use by parts Derivative of x and integral of cotx and then int(ln(sinx)) from 0 to π/2 will be -(π/2)ln2 using king's property and some basic evaluation of integrals and there is -ve sign in by part term so it will be (π/2)ln2 and initial part xlog(sinx) from 0 to π/2 would be zero so final answer would be (π/2)ln2..
Idk if you read the comments on your old videos but you're basically my math professor for calculus :)
@@English_shahriar1 stop advertising your yt channel in comments
This question was in my calc 1 final in 2021... I now know why this was so freaking hard.
You must be a math GOD. Just your skills with a marker are proof enough. 😮
U can also solve by parts. first xtanx = xcotx then by parts. integration from 0 to pi/2 ln(sinx) is -pi/2 ln2
I think I should add a few comments on the last parts, where you cancel a-1 directly and sub a=1 to the solution. There should be a limit instead of just sub a in directly, because what you did in the integral part w.r.t. x is an extension of the original integral, which has singularity when a=1 during partial fraction. At last when we sub a=1 to the solution, I think the continuity should be specified using convergence of original integral.
Beautiful
Fantastic!
It can be done by using definite integral property, a limit and the beautiful integral of 0 to π/2, ln(secx)
It's simple take it as integral of xcotx and then apply integration by parts And after that put the limiting values from 0---> π/2
Great Thank you teacher Please more like this *bP🖋️rP🖍️* ❤️
Good luck on the BMT2022, Man, do your best
Holy shritting crap, a *heart* from the legend him self, you don't know how much I learned from you, I didn't even take Calculus in school yet, yet here I am, solving with you, thank you so very much, and hopefully, *hopefully* , you get first place 💜💜💜💜💜💜 Edit: I now realize how funny this word "shritting" is 😂
That’s a really cool problem
Bravo!
I solved this question with the help of complex numbers. x over tan(x) equals to x times cot(x). And from here we find x times ln(sinx) minus ln(sinx) in the integral. sin(x) equal to (e^(2ix)-1)/2ie^(ix). İt's easy after that😉. İf it wasn't for BPRP, I wouldn't have such a great mindset. Thanks...
I too can fill a whiteboard with lots of integral formulae in red and black. Usually takes a lot of beer first. See you at the math comp!!
that's beautiful
This is so cool.
Anothe method to bring linearity is substituting 1/tanx with its trigonometric/parabolic equal which is ix(e^ix + e^-ix)/e^ix - e^-ix. It is easier to work from here. Another easier way is simplifying the equation to xcotx, it will bring a ln to integrate, but it is still easier to find methods past that than the method explained. But that is the beauty if differentiation and integration, there are many ways of getting the same answer, trig/hyperbolic functions are my choice, but whatever is easiest for each person.
Thank you appreciation from India
20:30 by putting a=0 here and having a>=0 won't we have an issue while inputting the limits in tan(au) when u approaches infinity. If a can be equal to zero won't we have a possible 0×infinity situation
You're right. Instead, we should consider that lim(a->0+) I(a) = lim(a->0+) (pi/2 * ln(a+1) + C) = C. We can see from definition of I that it is continuous and I(0) = 0, so C = 0 as well.
@@BrollyyLSSJ thanks so much, was wondering about this caveat too
@@asparkdeity8717 me too...
@Brollyy there you assume that a*u (with a ->0 and u->infinity) equals infinity, else I(a) isn’t continuous for a = 0. The problem with this limit is that a*u can be any (positive) real number or infinity. The definition of I(a) you used assumed that a was not equal to zero, so if you change that, you have to calculate everything again
@@alexandervanhaastrecht7957 You are right - if you really want to be rigorous, you need to show continuity from above at *a = 0* using the integral definition of *I(a).* Luckily, the integrand (call it *f_a(x)* ) can be continuously extended to *x = 0* via *f_a(0) = a.* The extended integrand converges uniformly to zero on *[0; 𝞹/2]* as *a -> 0^+,* proving continuity from above of *I(a)* at *a = 0*
Use a+b-x property of definite integrals it will be easier
I actually learned about the value of integral of lnsinx in class beforehand so it was just a matter of applying ILATE rule You first prove that integral of ln(sinx)=ln(sin2x)=ln(cosx).The value will come out to be π/2ln2. Then just convert the given integral into xcotx and apply ILATE rule
Love your videos :) Love your enthusiasm more!
Thank you!
So good
This seems like magic to me.
I have not learned any calculus yet but this is very interesting!! 10/10
Why so complicated? Here are the steps I took:- 1) Do integration by parts taking x as the first function and 1/tan x as the second function as 1/tan x is easy to integrate over 0 to pi/2. (~20 seconds) 2) Find integral of 1/tan x over 0 to pi/2 by doing u-substitution, u = sin x , you'll get integral of 1/tan x as log(sin x) in like 2 sub-steps. (~1 minute) 3) When we put value of integral 1/tan x in our Integration by parts equation of step 1, we get [x log(sin x) - integral (log(sin x) dx ] 0 to pi/2 Second term looks wild but It's easy to solve using properties of definite integrals (It was in my 12th grade NCERT book, I even remember the result). Int 0 to a F(A-x) = F(x) (~10 minutes) We get -pi/2 log(2) for the second term. Bam! We get pi/2 log(2) as the answer
pi/2 log(2) is about half the amount BPRP got.
@@HappyGardenOfLife I think they meant (pi/2)ln2
Good luck integrating ln(sin(x))
@@seroujghazarian6343 🤣
@@seroujghazarian6343 :tf:
beautiful
Dang this video owns
If we consider the integral as xcotx we can do it integration by parts and ans comes out to be π/2 ln2
Challenging! Good class and exercise!
Wonderful u used the famous feyymen trick at last
Integration by parts gives us -Int(ln(sin(x)),x=0..π/2)
We can convert the definite integral to integral tanX(pi/2-X) dX, Then use integration by Parts, we get 0 -integral log(cosX) dX. Which can be solved to get Pi/2 log2
Good job
Snap. Can't wait until I get to Calc III and can really understand what the multivariable stuff is all about.
Many integrals like this one can be handled more effectively and efficiently in the complex world by using contour integration
I applied the limit rule then I got tan(x)(pi/2-x) then I integrated tan(x)x using the power series expansion of tan(x)
The result is absolutely beautiful. The path to get there is horrific. I think I'll have mathmares (math nightmares) now.
That's crazy!
I can't imagine I really used to solve all this a few years back 😂
And now?
@@ShanBojack usually, you study all this rigorously in school just to realise it’s not used as much in the working world. He probably hasn’t touched this since graduating.
@@ShanBojack Keep using it or lose it. 😅😂
@@ShanBojack dude its actually worthless and not efficient at all to solve this by yourself...nobody will pay you for this nowadays. Of course its cool and stuff but in reality its just wasting time at the end of the day.
@@navjotsingh2251 cmon this stuff is beautiful , and if you dont do anything you dont need anything to do what you are doing and that would be nothing hence you need nothing to do nothing (in any field)
Bro we can directly apply by parts and we would end up with the common integral int(0toπ/2)ln(sinx) which is directly -π/2ln2
No need double integral (or derivation under integral sign) integration by part is ok here since 1/tan(x)=cos x/sin x=f'/f has an obvious antiderivative.
i dont understand anything from these videos but i watch bcs its fun :D
wow just wow, i love math
This can be solved very easily using the properties of definite inegrals and complex analysis to get π/2ln2
Geometrically it’s equivalent to the integral of the area as function of a of 1/(1+(a tan (x))^2) from o to pi/w evaluated at a=1!