![blackpenredpen](/img/default-banner.jpg)
- Видео 2 338
- Просмотров 207 533 106
blackpenredpen
США
Добавлен 29 дек 2012
I share the fun of solving math problems.
Check out my other channels "bprp calculus tutorials" or "bprp math basics" for math tutorials for your class.
Check out my other channels "bprp calculus tutorials" or "bprp math basics" for math tutorials for your class.
Can we use power series to evaluate the limit of a multi-variable function?
I came up with the limit of (e^x-e^y)/(x-y) as (x,y) goes to (0,0) and thought about using power series expansion for e^x and e^y to help us evaluate the limit. We know both series have infinitely many terms but do they have the same amount of terms that allow us to do the factoring in the video? If you know the answer, please let me know. Thank you!
Check out @bprpcalculusbasics for more calculus 3 tutorials!
----------------------------------------
Big thanks to my Patrons for the full-marathon support!
Ben D, Grant S, Mark M, Phillippe S. Michael Z, Nicole D. Camille E.
Nolan C. Jan P. Devun C. Stefan C.
💪 Support this channel and get my math notes by becoming a patron: www.patreon.com/blac...
Check out @bprpcalculusbasics for more calculus 3 tutorials!
----------------------------------------
Big thanks to my Patrons for the full-marathon support!
Ben D, Grant S, Mark M, Phillippe S. Michael Z, Nicole D. Camille E.
Nolan C. Jan P. Devun C. Stefan C.
💪 Support this channel and get my math notes by becoming a patron: www.patreon.com/blac...
Просмотров: 2 638
Видео
What is f''(x) in terms of dx/dy?
Просмотров 20 тыс.День назад
Try Brilliant with a 30-day free trial 👉 brilliant.org/blackpenredpen/ ( 20% off with this link!) You might noticed that f'(x)=dy/dx = 1/(dx/dy), but how about d^2y/dx^2 ? Let's work out this fun calculus question with implicit differentiation and the chain rule for derivatives. Big thanks to my Patrons for the full-marathon support! Ben D, Grant S, Mark M, Phillippe S. Michael Z, Nicole D. Cam...
limit of the quadratic formula as (a, b, c) goes to (0, 0, 0)
Просмотров 29 тыс.День назад
We will determine the limit of the quadratic formula as (a, b, c) goes to (0, 0, 0). This is going to be a fun one! Check out more limits of multivariable functions here: ruclips.net/video/8ek_v0zikP0/видео.htmlsi=G3zQpiev8fLzm48i Big thanks to my Patrons for the full-marathon support! Ben D, Grant S, Mark M, Phillippe S. Michael Z, Nicole D. Camille E. Nolan C. Jan P. 💪 Support this channel an...
The most fun way of solving this quartic equation
Просмотров 34 тыс.14 дней назад
Try Brilliant with a 30-day free trial 👉 brilliant.org/blackpenredpen/ ( 20% off with this link!) I created this quartic equation x^4-x^2-2x-1=0 and we will solve it by using the quadratic formula three times! Check out how we can factor x^4-x^2-2x-1 by using the double-cross method: ruclips.net/video/kwZiaKytsSQ/видео.html Big thanks to my Patrons for the full-marathon support! Ben D, Grant S,...
Can the quadratic formula still work if a, b, c are not constants?
Просмотров 81 тыс.14 дней назад
I was driving back home and wondered if the quadratic formula still works if a, b, c are not constants? In fact, what if a, b, and c are functions of x? I created an incredible example based on this: ruclips.net/video/j6ri-2S-hxU/видео.html Newton's method (introduction & example) ruclips.net/video/iVOsU4tnouk/видео.html Big thanks to my Patrons for the full-marathon support! Ben D, Grant S, Er...
I finally took the limit of the quadratic formula
Просмотров 69 тыс.21 день назад
Math for fun! Taking the limit of the quadratic formula as a goes to infinity. Enjoy! Related videos: I differentiated the quadratic formula: ruclips.net/video/JEcE-wDRMCk/видео.htmlsi=b8kAyZmnWJpUcD1l I integrated the quadratic formula: ruclips.net/video/J-p1tkaZKNg/видео.htmlsi=tJfByRfw41NpP0JV Solve 0x^2 bx c=0 by @MichaelPennMath ruclips.net/video/wTEYApUX-N8/видео.htmlsi=NqbM6M_sElo2vj2J B...
Math for fun: integral of ln(x) from 1 to ? is equal to 2
Просмотров 50 тыс.Месяц назад
Get started with a 30-day free trial on Brilliant: 👉brilliant.org/blackpenredpen/ ( 20% off with this link!) I want the area under the curve y=ln(x) from 1 to some number t to be 2, but how can we achieve this? Not only do we have to use calculus integration by parts, but we also need to use the Lambert W function to solve the resulting equation for us. 💪 Support this channel and get my math no...
Oxford MAT asks: sin(72 degrees)
Просмотров 135 тыс.Месяц назад
Get started with a 30-day free trial on Brilliant: 👉brilliant.org/blackpenredpen/ ( 20% off with this link!) We will evaluate the exact value of sin(72 degrees) via the sin(5 theta) formula. This question is from the University of Oxford Math Admission Test in 2022 www.maths.ox.ac.uk/system/files/attachments/test22.pdf Big thanks to my Patrons for the full-marathon support! Ben D, Grant S, Erik...
Meet the 94-year-old calculus teacher! @ycmathematicsphysicsandche5659
Просмотров 23 тыс.Месяц назад
Thanks to the RUclips algorithm, I discovered Mr. Feng, a 94-year-old calculus teacher, @ycmathematicsphysicsandche5659 . He said in his note that teaching makes him happy and makes his life meaningful, which I resonate with so much. I want to make a video to introduce him to you! Please consider subscribing to his channel and supporting what he does. Thank you. The solution to the derivative c...
First time solving an A-Level maths exam! (90 minutes, uncut)
Просмотров 162 тыс.2 месяца назад
I will be doing a British A-Level further maths paper on the spot for the first time! This paper contains mainly algebra and calculus. Topics include complex numbers, hyperbolic equations, exponential equations, trigonometric identities and equations, first-order linear differential equations, determinant and the inverse of a 3x3 matrix, and more! Was I able to solve all the questions within 90...
How to solve the three-circle problem from the 2022 GCSE math exam
Просмотров 78 тыс.2 месяца назад
Get started with a 30-day free trial on Brilliant: 👉brilliant.org/blackpenredpen/ ( 20% off with this link!) Here's the last question from the 2022 GCSE maths paper that made the news. We have three circles as shown and each radius is 4 cm. We have to find the area of the shaded region in the middle. I made a horrible mistake last time when I said the area of a circular sector is r*theta. The c...
Solutions to the 2023 AP Calc AB FRQ
Просмотров 32 тыс.2 месяца назад
Get started with a 30-day free trial on Brilliant: 👉brilliant.org/blackpenredpen/ ( 20% off with this link!) We will go over ALL the free response questions from the 2023 AP Calculus AB test to help you prepare for the upcoming AP exam. Topics included defined integral, average value of a function on an interval, change in position vs total distance traveled, implicit differentiation, related r...
You see nonlinear equations, they see linear algebra! (Harvard-MIT math tournament)
Просмотров 145 тыс.3 месяца назад
Get started with a 30-day free trial on Brilliant: 👉brilliant.org/blackpenredpen/ ( 20% off with this link!) This system of nonlinear equations is from the general round of the 2023 Harvard-MIT math tournament. www.hmmt.org/www/archive/271 I will present the linear algebra method I learned from their official solution to solve this system because I thought it was fascinating. It's from Harvard ...
Which is the worst math debate: 0^0, sqrt(1), 0.999...=1, or 12/3(4)?
Просмотров 287 тыс.3 месяца назад
These are the most debated math topics on the Internet but which one is the worst? (A) 0 to the 0th power=1 or undefined. No calculus limit here. (B) sqrt(1) = 1 or both -1?) (C) 0.999...=1 or not? (D) order of operations 12/3(4)=1 or 16 More than 28,000 viewers voted in my recent poll and now let's discuss what each debate is all about. 🛍 Shop my math t-shirt & hoodies: amzn.to/3qBeuw6 💪 Get m...
You wouldn’t expect this "quadratic" equation to have 6 solutions!
Просмотров 124 тыс.3 месяца назад
You wouldn’t expect this "quadratic" equation to have 6 solutions!
Using a logarithm property to make this equation easier!
Просмотров 91 тыс.4 месяца назад
Using a logarithm property to make this equation easier!
My first calculus 3 limit on YouTube
Просмотров 92 тыс.4 месяца назад
My first calculus 3 limit on RUclips
Believe in geometry, not squaring both sides!
Просмотров 268 тыс.6 месяцев назад
Believe in geometry, not squaring both sides!
Solution of the transcendental equation a^x+bx+c=0
Просмотров 157 тыс.6 месяцев назад
Solution of the transcendental equation a^x bx c=0
My failed attempts to the integral of sqrt(sin^2(x))
Просмотров 112 тыс.6 месяцев назад
My failed attempts to the integral of sqrt(sin^2(x))
Calculus teacher vs L'Hopital's rule students
Просмотров 88 тыс.7 месяцев назад
Calculus teacher vs L'Hopital's rule students
combining rational exponents, but using calculus,
Просмотров 98 тыс.7 месяцев назад
combining rational exponents, but using calculus,
easy derivative but it took me 32 minutes
Просмотров 182 тыс.7 месяцев назад
easy derivative but it took me 32 minutes
I couldn't solve x^x^x=2, so I solved x^x^(x+1)=2 instead
Просмотров 127 тыс.8 месяцев назад
I couldn't solve x^x^x=2, so I solved x^x^(x 1)=2 instead
all solutions to 2^x-3x-1=0 (transcendental equation)
Просмотров 142 тыс.8 месяцев назад
all solutions to 2^x-3x-1=0 (transcendental equation)
The step you highlighted is correct. If you write them as limits of partial sums, there are two independent limits creating a double sum - if this double sum converges, then adding restrictions to the parameters won't change the convergence. More directly, the function e^x is analytic in x, so we know that it has a convergent taylor series (same for e^y in terms of y), then writing them as a double sum, the double sum is independent of path (the only thing that matters is that the upper index approaches (inf,inf), regardless of path), which is a more general continuity than to pick a specific path as you did by declaring the upper index to go along the diagonal. In fact, by path independence, you could take either part of the index to be larger, and the remaining terms that don't have a partner to factor with in your factoring scheme will be negligible in the limit, same as they are in the taylor series of the original e^x (or e^y). The step that bothers me is the last one. Plugging in (x,y)=(0,0) is a statement of continuity. We can normally do this to taylor series, because they are already analytic and thus continuous. But the final series is, in a way, simply a renaming of the original function - does that make it a valid 2-variable taylor series? I don't know, but maybe it is known. I'm pretty sure it is valid here, but I don't know if it is always a given or if it requires an exchange of limits I'm too lazy to verify right now. Can anyone give me an overview on multivariable taylor series?
As others have pointed out, I think this method works fine, as long as the series you are using are absolutely convergent, which they are here.
The expression is not defined for x=y so the limit doesn't exist
Quizás se pueda demostrar con la definición epsilon - delta para límites de varías variables.
x^n - y^n can always be factored as (x-y)*(x^(n-1)+x^(n-2)*y+...+y^(n-1)). So you do this thing here and get 1 as the result
Is there an integral you could make out of this instead? I mean I think of Riemann sums when I see 0+0+0+0+0+0+yada yada
The Blackpenredpen Conjecture
I would argue as it’s the same expansion for the same base function, it would be the same number of terms, even if it is infinitely large. I guess it’s one of those ones where it depends on how it’s been defined and the convention
Lemme try. Let n(r) be the rth term of the infinite sum. From my observation, n2=(x+y) (2 terms), n3=x^2+xy+y^2 (3 terms)… lim n →0(n^p) (p=a positive integer) will get closer to true 0 as p gets bigger If x=y, n3=lim x→0 (x^2+x*x+x^2)=3x^2… Basically, n(r)=lim x→0 r*x^(r-1) This will also approach true 0 as r gets bigger, due to the decreasing rate of exponentiation when x<0 greater than the increasing rate of multiplication It is also noted that each term n(r) is divided by a constant r! Hence, the nth term will get closer to true 0 than the (n-1)th, (n-2)th, and basically every term before it So, n(∞)=lim (n,x)→(∞,0) (n*x^(n-1)/n!)=(x^(n-1)/(n-1)!), which is a term incredibly close to absolute 0 Hence, we can say with confidence that all the terms after n1 are 0. Thus, the sum=1+0+0+0+…… *=1*
I challenge you to complete the multivariable limit from hell: lim x-> 0, y-> 0 (sin(xy)/xy)^[(1)/(1-cos(xy))] LaTeX Code: \left(\frac{\sin(xy)}{xy} ight)^{\frac{1}{1-\cos(xy)}}
Both sets are obviously countable, so I guess for any number you can combine and factor out 2 members of sets. Don’t see an issue.
This step has to be legitimate. I mean, yes, those are two different sums, however, every addendum is numerated, so for every nth addendum from the first sum we can always find respective nth addendum from the second sum. In other words, for every 1/(n-1)! × x^n we can find 1/(n-1)! × y^n no matter how far n goes into infinity.
As an engineer, I'd "linearize" the problem by considering lim/x-->0 e^x = 1 + x, right up front. In other words, you consider the linear approximation of e^x in the region where x = 0. Then you have (1 + x - 1 - y)/ (x-y) = 1. I'll leave it to the mathematicians to show this is mathematically correct ;-)
That's an interesting way to solve this problem, but I'm not entirely sure what the rules for infinite series are. Related to multivariable limits, is it normal to use L'Hopital? To explain my reasoning, any f(x,y) can be written as f(x(t),y(t)), right? That's a single variable function now, so if you have f(x(t),y(t))/g(x(t),y(t))=0/0 then you should be able to use L'Hopital and the chain rule. I will write x(t) as x and x'(t) as x' for simplicity. This is how I solve this problem: (e^x-e^y)/(x-y)=(x'e^x-y'e^y)/(x'-y')=(x'-y')/(x'-y') for this, we have 2 cases if x'-y'=0, which means x=y+c, but since x and y are 0, c must be 0, so you really just have x=y, which we can see will give the limit 1 else fraction simplifies to 1 I haven't defined the functions x(t) nor y(t) or their derivatives, so I haven't added any restrictions to the problem by doing this. Do you have anything similar on the channel?
i think you ran into the 0 times infinity because i think the zero repeating infinatly many times and you just considered it as 0, love your videos sir hope to get your T-shirt <3
Also we also gotta show that the terms in the bracket have a common factor of x-y ie x⁴-y⁴, x⁵-y⁵, upto infinitely many terms. This can be shown by mathematical induction
On the path of x=y, the value is indeterminate unless you impose a value of 1 based on limits.
Stop doing easy videos, even tho I’m a kid
To be continued?
You are enjoying the maths in better way, You are very energetic dude🤟🫶
I am afraid there is some problem for this approach. The reason behind is that there is NO correlation between x and y. The limit will totally different if you select different path for x and y to move toward (0,0). A path x=y will come up with the answer 1, while x=10y path will come up to a limit e^10/9. 🤔
We can only tell the answer is correct if and only if we set a criteria x tend to y and y->0.
x and y are independent, yet this limit will always be 1, it's that simple. Yeah, in lots of cases it doesn't exist but here it does
Thanks PROF UR VIDEOS are the BEST
i havent done multivariable calc but cant you solve it using a substitution? lim x,y --> 0,0 of e^y(e^(x-y) - 1)/(x-y) and if the normal rules are still valid, limit of a product = product of a limit so we can isolate the e^y we factored and for the remaining limit make a substitution of x-y = t, and as x and y approach the origin, t will approach 0 [lim (x,y)--> (0,0) of e^y ] * [lim t-->0 of (e^t - 1)/t] which is just e^0 * 1 since the 2nd limit is just a standard limit so the final answer still ends up to be 1 but since i dont know multivariable calc i dont know if the substitution i made i valid or not
Does cancelling our the x-y not just mean you're dividing by zero? Similar to the 1=2 'proof'?
they arent equal. its a limit not a computation
Why not make the substitution y=x-z, Then you can turn it into a single-variable limit.
The value of the limit is indeed 1 (proof by 3D desmos lol). However, by this reasoning, we use Taylor series for two analytic functions, so we do: lim (x,y) -> (0,0) then (lim n -> ∞ of "first sum" + lim m -> ∞ of "second sum"). (I will put A="first sum" and B = "second sum" for the rest) So the question is whether we can do: lim n -> ∞ of A + lim m -> ∞ of B = lim n-> ∞ of A+B. And seen like that the answer is obvious: lim n -> ∞ of (A+B) = lim n -> ∞ of A + lim n -> ∞ of B. (Indeed, we can use the linearity of the limit because the two limits exist because exp is an analytic function). So the reasoning is good and the limit is equal to 1 □. Edit: I just proved that lim of A + lim of B = equal lim of (A+B) in this case. Moreover we can rearrange the terms (thus factor like you’ve done) because *exp converges absolutely*
Isn’t the possibility of constructing bijection between terms of the infinite series a proof of them having the same „number” of terms? I would this bijection as „swap x with y” and the reverse one as „swap y with x”.
maybe, the life is better in the U world....
y=x+h lim (h -> 0) (e^x - e^(x+h))/(x-(x+h)) = lim (h -> 0) (e^(x+h) - e^x)/h = e^x lim (x -> 0) e^x = 1
as long as x and y are never equal this holds true, even if x and y have a big difference, its still around 1
What about highlighting the infinite nature of the series explicitly: Lim(x,y)->(0,0) (e^x- e^y )/ (x-y) Lim(x,y)->(0,0) limN->infty Sum(n=0)N (x^n - y^n)/ n!(x-y) Now I believe your question is : what allows us to set a single N to both expansions instead of doing two limits with M and N for x and y respectively? I believe this becomes clear for a more general case if we use expansion of multivariete functions! Set f(x,y) = e^x -e^y. We now do (taking the limits implicitly) f(x,y) = sum(nx)N sum(ny)M 1/(nx!ny!)f^(nx,ny) x^nx y^ny. We now do l=nx+ny and do L=M+N. We get sum(l=0)L sum(n=0)l f^(l-n,n) x^l-n y^n Where now we only take the limit of M+N=L . We basically use a Riemann product of series. Now this function in particular is such that mixed derivatives yield 0. Therefore we have the series: Sum(n=0)infty 1/n! (x^n-y^n).
Also anyone who sees this, let me know if there's anything wrong and sorry for the notation here!
@@carlosp.2898imo this is good. However, we also need to prove that we can rearrange the terms in the Taylor serie (because bprp factors the (x-y)) but this can be proven easily because the sum converges absolutely
@@matonphare you mean in the Riemann product right? Yes, i kinda skipped that and took it for granted because of what you said!
@@carlosp.2898 yeah I meant that my bad
By MVT there exist c between x and y such that (e^x-e^y)/(x-y)=e^c. Hence if (x,y) approach (0,0) then c will approach 0 (by squeeze). Thus the limit is simply e^c as c approaches 0, that is 1
Since the expansion of e^x is convergent, I think your method is valid
I feel it's wrong. If we approach along a path where x goes to 0 much faster than y, then we only need to expand till like 3 terms of e^x to get a good approximation and many terms of y, but if they have unequal number of terms it doesn't work But I am not sure
While that’s true for an approximation, the series expansion is supposed to act more as a limit to infinity approaching the exact value, so I think that the number of terms should both always be approaching infinity since we’re talking about an exact value which is only obtained at the limit to infinity. At least that is what I think but it does always get tricky with infinite series so you might be right
Man, you hit the nail on the head! This series isn't converge as x and y are not related!
The point here is not to use Taylor expansion, but truncated expansion until order 3. Then, due the guaranteed conditions of the remaining functions you'll be able to continue.
Can you explain what that means please? What other conditions?
I think this is fine as the series both converge absolutely, so we are free to rearrange them without changing their value. Another way to look at it is to change variables to u=x-y, v=y. Then the expression becomes lim e^v (e^u-1)/u, decoupling the variables into two separate factors. The limit of the e^v term is 1, and limit of the (e^u-1)/u term is also 1 by definition of the derivative of e^u. Using algebra of limits gives the result.
Is breaking limits in the multiples also valid in multiple variables?
I would suggest transferring this problem into complex analysis to make things simpler where we have analytic functions of multiple complex variables meaning that we can use the techniques of complex analysis to simplify things. I think we would end up with a function with coefficients being other complex analytic functions, which should probably be well defined.
Seems legit to be, would look nicer if you would have used sigma notation
With your videos, students all over the world don't need uni professors anymore!!!!
I admire your honesty and modesty,"i am not sure if this is correct"💯👏👏
BPRP is absolutely good, but here is nothing related to honesty at all. Math is a thing that needs rigorous proof -- mathematician can't tell anything that without a solid ground. If they do, time will tell. For this video, a double variable limit will be the catch here since two variable are not necessary the same. That means they can "float" in 2 completely different dimension. We can't jump to a conclusion that the series is converge as x is nothing related to y with both of them tend to zero, though. If the question changed to "x->y and y->0", the answer should be correct! But this is not the question asked! So, I doubt! 🤔
Nice video! Now I can understand why (e^x - 1)/x when x approaches to 0 equals 1 without L'hopital rule
That's just the limit definition of the derivative of e^x evaluated at 0
(sorry for my bad English and sorry for my bad math, im still at school) i think this step is correct because we have “equals” things: in left of side ”x”, in right side “y” and if we know how look expression in left side we know how it look right side and they close to be equals, because of that we have 1/n!* (x-y)^n, i think its very easy to imagine please tell me if i have error, i really want understand why we have problem in this step
Idk if im correct, but i think we can. Since, we do something like this only while proving the euler's formula of exp(iθ) = cos(θ) + isin(θ)
I do think you are correct here, because the power series (by definition) would have infinite terms, no? And every term is going to have a factor of (x - y). Whether you do this over 3 terms, 30 terms, or 300 terms, there will always exist a factor of (x - y), which means that it can be factored out. I can't say with absolute certainty that it is correct, but I also cannot think of any reason why it would not be correct. Interesting video!
Yes, but you can can expand both to a different number of terms, so I don't think it's correct
@@thatapollo7773 As long as the number of terms it is expanded for it clearly defined to be equal, then I think it is still valid.
@@terryhand5871 Yes, but the point here is --- the question didn't express clearly that x tend to y. This means x and y are two independent plane which different limits will be obtained through different path toward the (0,0) point. We can't assume this unless the questioned stated explicitly x tend to y. Or, in another way round, x and y approach zero on the straight line x=y.
I think is for sure right
This is super cool whether it’s right or wrong, I love your videos dude
Thanks!
Nice way.
can you make a video on very fun derivatives? Really love your way of teaching !😁
Sleep
Can you make video about Hartley transform or some unusual integral transforms