why the area under the tangent line of 1/x is special

Подписаться 1,3 млн

Просмотров 312 тыс.

50% 1

Visit 👉 brilliant.org/... to learn more calculus (20% off with this link)!
Learn why the area under the tangent line of y=1/x is very special. We need to know how to use the power rule to take the derivative of 1/x and how to write the equation of a tangent line. Then we just need the area of a triangle formula, which is just 1/2*base*height. This is a very classic calculus problem and it's great for Calculus 1 or AP Calculus AB students. Subscribe for more math for fun videos 👉 ‪@blackpenredpen‬
#calculus #blackpenredpen #apcalculus #chanllenge
---------------------------------------------------------------------------------------------------
**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
---------------------------------------------------------------------------------------------------
💪 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

Опубликовано:

26 сен 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 420

@blackpenredpen 2 года назад

Learn calculus in a Brilliant interactive way 👉 brilliant.org/blackpenredpen/ You can help this channel a lot by simply clicking the link and checking them out. Thank you 😃

@Mnemonic-X 2 года назад

Do you know why Einstein was really stupid? I know why.

@CONEJOI666 2 года назад

Thanks to teach this exersice, my teacher of calculus when appear this exersice, explained so horrible and i cannot understand nothing, but seeing ur video i have clear the method

@ahmadtim2537 2 года назад

For the function f(x)=m/x^n, the area under the tangent line when x=a is (m×(n+1)^2)/(2n×a^(n-1)) so it that's why the area under the tangent line of f(x)=1/x is special. Edit: thanks for being the best math guy on youtube.

@ryanb1874 2 года назад

your serious, all angles of tangent will produce same area?

@StichyWichy21 2 года назад

General formula for area for a function f(a): A(a) = af - ½a²f' - f²/2f' From this it can be found the only functions for which area is constant are of the form f(x) = Ax + B (which is trivial), and f(x) = A/x

@davidseed2939 2 года назад

thank you very much. usually my comments are solutions. In this case, my comments were questions. It makes it more interesting (and less tiring!)

@peterclark5244 2 года назад

I was going to try and puzzle this out, thanks!

@saif_saleh_ 2 года назад

What is this called? I'd like to learn more about its derivation, but I can't find anything online.

@mahmoudfathy2074 2 года назад

@@saif_saleh_ just do as shown in this video. But generalizing f(x), it's not rigorous i guess but it will end up with this formula. Start: m = f'(a) p = (a, f(a))

@virusweb7080 2 года назад

@@mahmoudfathy2074 It is rigorous actually. af is the rectangle which I'm sure you can visualize. There are 2 triangles left to add. One of which will have its base as a. Since the slope is f', the other leg has to be of length a*f' and same reasoning for the other triangle.

@ewancox7017 2 года назад

I'm jealous of people who have him as his teacher

@blackpenredpen 2 года назад

My students might think otherwise 😆

@ewancox7017 2 года назад

@@blackpenredpen haha, I guess I can't really complain when I go to a Maths based College in England

@combcomclrlsr 2 года назад

You mean envious. Jealousy is fear of losing someone to another.

@JP-lz3vk 2 года назад

@@blackpenredpen I assume the students want to learn from a passionate teacher like yourself and are not forced to be there. But the best teachers demand the best from the their students and since you have a RU-vid channel, no-one can say they were not warned....

@RS-jq4oc 2 года назад

@@JP-lz3vk well said.

@leonardsmith9870 2 года назад

I nearly failed math in high school but you've really re-ignited my passion for math at the ripe age of 31. I've recently upgraded my high school math subjects with high 90s in order to get into university! Thank you for what you do and getting me back into it!

@Roescoe 2 года назад

This man is subbed to Animenz. Good taste :D

@grgfrg7 2 года назад

Thanks for posting my friend I am also 31 and thinking about going back to university to study a different field. Good to see others in my situation making it happen

@Smallpriest 2 года назад

@@Roescoe He's also subbed to Pan Piano, what a man of culture

@gaetanramos7903 2 года назад

Good luck with university man, that's awesome.

@jakobr_ 2 года назад

This probably has some connection to the fact that the rectangle with one corner at the origin and the opposite corner on y=1/x has the same area=1 no matter what point on the line you choose. The connection to the “area under tangent line = 2” is easy to see geometrically once we’ve shown the tangent’s intercepts are twice the coordinates of the point on the curve. Can anyone think of a non-calculus way of getting to that point?

@jeremydavis3631 2 года назад

Here's one (it helps to draw the figure as you read about it): Consider the rectangle OBAC O(0,0), A(a,1/a), B(a, 0) and C(0, 1/a). Clearly, the area of this rectangle is 1. Now split the rectangle along its diagonal OA, forming two right triangles: OAB and OAC. Reflect triangle OAB across line AB and segment OAC across segment AC. This forms a new triangle ODE with D(2a,0) and E(0,2/a) and with exactly twice the original area, i.e. 2. To prove that segment DE actually passes through A: Angle BAC is a right angle, and the reflections preserved angles. Therefore, angles DAB and EAC are complementary, and the the three angles altogether are supplementary. So lines AD and AE are parallel. Since they both contain A, they are the same line. To prove that line DE is indeed tangent to y=1/x: We already know that A is on the curve, and given the shape of the curve and the negative slope of line DE, in order for the line and curve not to be tangent, there must be a second intersection point, Z, such that Z≠A. Suppose we find such a point at (z,1/z). Then line EZ must contain point D(2a,0). The equation of line EZ is y=(2/a-1/z)/(-z)x+2/a. Plugging in y=0 leads to x=2z/(a(2/a-1/z)). By our supposition, x=2a. So, solving for z: 2a=2z/(a(2/a-1/z)) a=z/(2-a/z) 2a-a^2/z=z 2az-a^2=z^2 z^2-2az-a^2=0 z=(2a±√(4a^2-4a^2))/2 z=2a/2 z=a But if z=a, then Z=A, which is a contradiction. Therefore, A is the only intersection of line DE and the curve defined by y=1/x, which are therefore tangent. Therefore, the initial construction was valid, and the area of the triangle must always be 2. Q.E.D. (That was a fun challenge. 😊)

@GopaiCheems 2 года назад

Wow, wonderful observation! Indeed, that area of rectangle being 1 would come to use. We draw the tangent at A(a,1/a). Then we draw the rectangle with O(0,0) and (a,1/a) as opposite vertices. We know this area is 1. Let the other two vertices be B,C. Now we find the tangent equation. The slope is -1/a^2 hence the equation is (y-1/a) = -(x-a)/a^2. Clearly, the x-intercept (y=0) = 2a Hence done, as you pointed out how the intercepts being twice of the coordinates of the point are enough. But for completeness, we draw AB, OA and AC to split the triangle in 4 regions. Two of these are equal to each other, as are the other two. Hence area of rectangle = 1/2(Area of triangle) = 1. For a non-calculus derivation of the slope equation, we write y = mx + c xy = 1 Hence x(mx + c) = 1 => mx^2 + cx -1 = 0 has only one solution. So discriminant must be zero. Hence c^2 = -4m. Now substituting x=a, y=1/a finishes.

@GopaiCheems 2 года назад

@@jeremydavis3631 ah, identical I wonder if we can completely dispose with even coordinates and solve a purely geometric problem in terms of hyperbolas

@oioficina3043 2 года назад

youtuuu.tokyo/Dv1PhhYxWho

@bachirblackers7299 2 года назад

Jacob Roggy you may use the same point (a,1/a) to be the intersecting point of major and minor axes of some Ellipse, same ellipse is kissing both x axes and you axes yes he must have area of Pi ,

@OCTAGRAM 2 года назад

Students are often not taught properly hyperbolic functions, hyperbolic rotation and pseudo-Euclidian geometry. Circle stays intact after rotation by any angle. Similarly hyperbola stays intact after hyperbolic rotation by any angle. I can apply hyperbolic rotation to transform any tangent triangle into a single well-known one, for which I already know the area. A nice property of hyperbolic rotation is that from the Euclidian interpretation area stays the same. Because Euclidian and pseudo-Euclidian areas are the same. Distances are not the same, areas are. Thus, hyperbolically rotated triangle have the same area as well-known one. Hyperbolic rotation can be viewed as area-preserving affine transformation. Yet another reason for constant area. There is a similar thing in higher dimension. Take an infinite cone and a hyperboloid inside with asymptotes matching infinite cone walls. Now you can draw tangent plane to any point of hyperboloid, and for the same reasons the volume will be the same. Using cone axis rotation and hyperbolic rotation we can move any hyperboloid point to a single well-known point with well-known volume. In terms of special relativity, Lorentz transformation keep the light cone intact (although lightwave length changes, but it has no impact on our task). Lorentz transformations also keep intact the spacelike surface of all events equally distant from an origin. This surface is hyperboloid and a light cone is a degenerated case of hyperboloid. Back to 2D, there is one more thing you can do. Take two hyperbolas with same asymptotes. Take tangent line to one hyperbola and intersect with another hyperbola. You'll enclose a constant area. It works very similar to two concentric circles and a chord that is tangent to the inner circle. In Euclidian space we just cannot degenerate concentric circles into something as remarkable as a light cone. Yet another idea for 2D: find an ellipse that has axes same as hyperbola asymptotes, and is tangent to hyperbola in a chosen point. This time we are relying on affine transformation properties.

@snacke4500 2 года назад

Got this on a test not to long ago! I happened to solve the question and the answer blew my mind! One of my favourite questions!

@euva209 2 года назад

Simple but thought-provoking. It can also be shown that for any function y=C/x, the area of the triangle will be equal to 2C. If you start with y = Cx^n, it can be shown by using the same approach that the only way the area is independent of the tangent point (x1,y1) is if n= -1.

@slinkyslider 2 года назад

4:01 when I am presenting in front of the class

@sysyphenf8ewtfr603 2 года назад

4:12 - 4:19 it seems you forgot to cut this out in editing lol

@NotBroihon 2 года назад

1/x is the only function that works. I did the math for f(x) = 1/x^n (with n > 0) and the resulting area A is: A = (1/2)(n + 2 + 1/n) * a(1 - n) where a is the x coordinate of the tangent point. As you can see only for n = 1 (aka 1/x) this result is independent of a. For 1/x^2 A is 9/(4a), 1/x^3 A is 8/(3a^2) etc. I got the formular for A by integrating the tangent function t(x) from 0 to the root of t(x). The generalized tangent function is: t(x) = -n / (a^(n + 1)) * x + (1 + n)/a^n And the root is a/n + a. t(x) can be derived from f(a) and f'(a). Edit: it's just a triangle, no need to integrate the tangent function lmao Result is the same though.

@inexistente8622 2 года назад

I always love to see math extended to a "n-limit" lul. Its so nice to see a formula that surfices a value you can choose. (If i said anything off or didnt understand something properly, pardon my english lmao.)

@decare696 2 года назад

I did the general case for any function (amounted to solving a differential equation) and the only solutions are f(x) = ax+b and f(x) = c/x. The former isn't very interesting (kind of self-evident), so 1/x and scaled versions are really the only functions with this property.

@inexistente8622 2 года назад

@@decare696 Im really curious to know what you meant by "any function". Im still learning Diferential Equations, so its not like i could grasp whatever method you used X). Still, sounds marvelous!

@pedrosso0 2 года назад

@@decare696 I did general case and got the equation (y/y')^2-x^2 y''/2=0 But I don't know how to solve for y

@_Blazing_Inferno_ 2 года назад

It looks like you’re missing a ^, where it would be a^(1-n) rather than a(1-n), unless I’m mistaken. I was confused after reading the equation the first few times.

@volodymyrgandzhuk361 2 года назад

To generalize, if we consider the hyperbola y=(ax+b)/(cx+d) (of course, c≠0 and ad-bc≠0, otherwise it would be a line and not a hyperbola), then the area of the triangle formed by its asymptotes and any tangent is (2|ad-bc|)/(c²).

@mikeschieffer7899 2 года назад

At 5:26 when y=1/x^2 the area of the triangle is not constant. The area is 9/(4a) where a is the x coordinate of the tangent point.

@seroujghazarian6343 2 года назад

the general formula of the area of the triangle under the tangents to y=1/x^n at (a,1/a^n) is ((n+1)^2)/(2n×a^(n-1))

@josueramirez7247 2 года назад

I really enjoyed this demonstration. These days, I don’t get to do a lot of math exercises so I really appreciate the challenge at the end.

@AdoNir 2 года назад

Stopped at 0:23, took 5 minutes for myself to solve it, got the answer 2 for any tangent point, got amazed, still watched till the end because you deserve it!

@blackpenredpen 2 года назад

Thanks!

@tomasgalambos3115 2 года назад

Is anyone else as impressed as I am over how quickly and smoothly he switches markers? O.O

@bingcheng8419 2 года назад

1/x is the only function. Here is proof: Assume 'a' as parameter, and points are (2a,0) and (0,2/a), and area of triangle 2. The line function is y=-1/a^2x+2/a . If x is any increasing function of a: x(a), so y=-1/a^2*x(a)+2/a. This parametric curve should have a slop of -1/a^2. Solving the differential equation for x(a), get x(a)=a

@CoDMunichTV 2 года назад

well, what about 2/x or 3/x. they all have the same size triangle formed by any tangent.

@jafetguzman5012 2 года назад

@@CoDMunichTV 2/x and 3/x are all 1/x. Essentually, you are taking the parent funciton (1/x) and multiplying it by a constant (2 or 3).

@martyguild 2 года назад

@@jafetguzman5012 i mean yeah, but that means 1/x is not the only function: a/x for all a in the reals would account for every function satisfying this property.

@jafetguzman5012 2 года назад

@@martyguild I don't disagree. I assumed that's what 1/x meant, but in less mathy words.

@gaetanbouthors 2 года назад

@@jafetguzman5012 there are other fucntions that could work, this is only proof for triangles with an area of 2. 2/x and 3/x do work but with different areas, however it isn't what was implied by the original comment, and isn't obvious either

@lavy9740 2 года назад

Hey man, I just wanna say that you are my favorite math guy on the internet. You're energy and love for mathematics is infectious and makes me proud to be studying the magic of arithmetic

@talkgb 2 года назад

to find similar functions, we look for a general length and width of the triangle which is nicely just the respective x and y intercepts of the tangent. therefore, for length, we solve f’(a) (x - a) + f(a) = 0 for x and get x = a - f(a)/f’(a). for width, we solve y = f’(a)(0 - a) + f(a) and get y = f(a) - af’(a). therefore the area of the triangle is A = (a - f(a)/f’(a))(f(a) - af’(a))/2. expanding this, we get A = [2af(a) - f(a)^2/f’(a) - a^2f’(a)]/2. turning into quadratic for specific case A = 2 gives a^2f’(a)^2 + (4 - 2af(a))f’(a) + f(a)^2 = 0 isolating f’(a), we get f’(a) = [af(a) - 2 +/- 2 sqrt(1 - af(a))]/a^2 which looks better as y’ = (xy - 2 +/- 2 sqrt(1 - xy))/x^2 sadly we can’t solve this besides test functions but we can confirm that only polynomials that are degree -1 monomials could work for Y (or technically 0, but that can be trivially ruled out) because otherwise the square root will remain, which cannot be in the derivative of a polynomial.

@lazarjovanovic3642 2 года назад

We had similar question on the admission exam for the college back in 2015. The function was exp(-x), and the question was: "If the tangent line intersects the axes in points A & B, and the coordinate start is denoted with 0, what is the maximum area of the triangle ABO?"

@felixmarquezy6772 2 года назад

I love your videos thanks for helping me to progress faster in math !

@blackpenredpen 2 года назад

Happy to help!

@SuperMtheory 2 года назад

Your video inspired me to write a Desmos demonstration for my students. Thanks!

@GopaiCheems 2 года назад

This is definitely a property of rectangular hyperbolas, but would it work for hyperbolas in general?

@RS-jq4oc 2 года назад

Could you link to a source? I don't know what rectangular hyperbolas are

@GopaiCheems 2 года назад

@@RS-jq4oc do you know that there are a pair of lines which hyperbolas never meet, but come really close to? A rectangular hyperbola is one for which these two lines (called asymptotes) are perpendicular to each other. In this case (y = 1/x), the X axis and Y axis are the two asymptotes.

@OCTAGRAM 2 года назад

Yes, it will. You can apply affine transformation to general hyperbola to obtain rectangular one, for which you already know everything. Affine transformation scales area by constant factor, so equal areas remain equal. Tangent lines remain tangent.

@GopaiCheems 2 года назад

@@OCTAGRAM ah, wonderful!

@BigDBrian 2 года назад

if it's given that it's a constant, it's very easy to see it has to be 2, because if you just take the diagonal at (1,1) then it's the area of a 1x1 square and two 1x1 triangles

@SteveAcomb 2 года назад

The real fun bit is when you form a solid by rotating this curve around the x axis and get a cone with finite internal volume but infinite surface area; an endless funnel you could fill with paint but never paint

@plazmyx5998 2 года назад

ooh that's nice

@ElPibePi 2 года назад

All the functions that fulfill the constant area restriction must satisfy that: -[y-y'*x]^2/(2*y')=A where A is a constant (the area) This comes from A=x*Y/2 In the tangent line equation, you have that Y-y = y'(x-a) So, Y=y'*x+y-y'*a Y=m*x+b. Therefore m=y'; b=y-y'*a. Then we have to find x and Y Y=0=m*x+b x=-b/m And Y=m*0+b Y=b. Replacing: A=(-b/m)(b)/2=-b^2/(2*m) A=-[y-y'*x]^2/(2*y')

@cmilkau 2 года назад

The union of all such triangles with fixed area A covers exactly the area under A/(2x) and the hypotenuse midpoint traces that function.

@andrewbeck9303 2 года назад

Wow this was so cool!!

@blackpenredpen 2 года назад

Thanks!

@bonbondojoe1522 2 года назад

General formula for f(x) = x^n: A(a) = (n + 1)²/(2na^(n - 1))

@blackpenredpen 2 года назад

😮!!

@zeroplays9915 7 месяцев назад

xy=c^2 (rectangular hyperbola) and let point be (ct, c/t) and then slope of tangent is dy/dx= (dy/dt) / (dx/dt) = -1/t^2 giving us equation of tangent as x + t^2 y = 2ct. Then put c=1 and get all 3 coordinates (by putting x=0 and then y=0) then apply area of triangle matrix to get A=2

@lunstee 2 года назад

I'd approach this in a manner following your early hint that it doesn't matter where on the curve you take a tangent. Note that the curve y=1/x is unchanged if you do a coordinate transform of (x,y) -> (x*A,y/A) : this is a horizontal scaling, together with a vertical scaling such that the two cancel each other and put the resulting curve right on top of the original curve. Not only is the curve invariant under this transform, so is the area under the tangent triangle. We're stretching the triangle by a factor of A in one axis, and shrinking it by the same factor in the other axis to no net effect on area. Since this transform doesn't affect the area of the triangle, we can transform the tangent point to (1,1). By symmetry of x and y, this is half of a 2x2 square, with area 2. The other approach to this is to note that the slope of the tangent, the derivative of 1/x is -1/x^2, while the slope of the line from the origin to (x,y) is y/x=(1/x) /x = 1/x^2, or the negative of the tangent line slope. This means the tangent line is a horizontal or vertical reflection of the line from origin to (x,y). If we draw vertical and horizontal lines from (x,y) to the axis, we partition the triangle into four smaller congruent triangles, two of which form a box of size x by y, with area 1. The total of the four small triangles is thus area 2. You can also avoid invoking calculus to get to this conclusion of the two slopes being negative of each other by following the initial invariance argument instead.

@silksongreactions 2 года назад

The Y-intercept really got to him

@blackpenredpen 2 года назад

😆

@olz6928 2 года назад

In my calculus class I also proved that the middle point of the tangent line (from y-axis to x-axis) is always on the functio.

@Starcanum- 2 года назад

Another cool observation one can make is that (a,1/a) is the midpoint of the line segment between the axes intercepts. So a hyperbola could be seen as the set of midpoints of hypotenuses of right triangles of a given area, with their legs lying on two intersecting lines. If we try to change the angle between the lines, we don't need to separately prove that the resulting locus will be a hyperbola too as it's just one linear transformation away from being the y=1/x, and all the tangent lines and triangle measurements scale accordingly no matter where they are. But if one likes to indulge in overcalculation of obvious things like I do, we could.. Nevermind, I did it but not going to post it as I got more and more lazy with formatting and it was hideous to begin with. It did cancel out in the end, which was mildly satisfying but then I knew that was always going to happen.

@sk8erJG95 2 года назад

In general, if we have a curve f(x), then tangent line at x=a is given by y = f'(a)(x - a) + f(a), so the intercepts are given by y = f(a) - af'(a) and x = a - f(a)/f'(a). So the area of this triangle is A(a) = (1/2)(f(a) - af'(a))(a - f(a)/f'(a)) = (1/2)(af(a) - a^2f'(a) - f^2(a) + af(a)) = (1/2)(a[2f(a) - af'(a)] - f^2(a)). For this to be independent of a, we would need 2f(a) - af'(a) = 0 for all a, meaning 2f(a) = af'(a). Writing this as a differential equation gives y' = (2/x)y which can be solved via separation of variables: (1/y)dy = (2/x)dx ln(y) = ln(2/x) + C y = K*(2/x). So this shows the only functions with this property are multiples of f(x) = 1/x.

@MJ-np8kg 2 года назад

4:06 , him considering re-recording the video is just epic at this moment 😅

@CoDMunichTV 2 года назад

i had this excact problem as a project in university but i did it using the taylor-polynomial. and then i made an equation for finding any function with the same properties, so any function where it does not matter where you put the tangent line and the area of the trianlge will always be the same. it was a very interesting problem thinking back now

@pwmiles56 2 года назад

That's an interesting question at the end. The answer is no, only y=1/x solves this particular problem i.e. A=2. A way to see this is to make lots of triangles, which will have their hypotenauses between points (2x,0) and (0,2/x). The hypotenauses border a curve called the envelope. So the envelope is y=1/x, that's the only solution. (You could put A=2*a^2 but that would just give y=a^2/x)

@Green_Eclipse 2 года назад

Yeah I found the same thing. The area being constant (rather than specifically 2) has 3 cases: y''=0 (line) y/y'-x=0 (subset of case 1) y/y'+x=0 (y=c/x) Which is either a line or proportional to 1/x. (I assumed that the tangent line just went to the x and y axis there could be some weird curve with self intersections but IDK how to compute that generically)

@azzteke 2 года назад

hypotenUses!!

@jakobr_ 2 года назад

Straight lines work as well. All tangent lines of a straight line are the same line, and therefore have the same area underneath. With an exception given to perfectly horizontal or vertical lines.

@oioficina3043 2 года назад

youtuuu.tokyo/mE0pkxJNojS

@jpgm11 2 года назад

With the point-slope equation we get: y=f'(a)*[x-a]+f(a) Whose x-intercept (y=0) is x=a-f(a)/f'(a) and y-intercept (x=0) is y=f(a)-a*f'(a) so the area under the triangle formed by the tangent line is: A=1/2*[a-f(a)/f'(a)]*[f(a)-a*f'(a)] =1/2*[2*a*f(a)-a²f'(a)-f²(a)/f'(a)] So to get a constant area, f must satisfy the differential equation. 2*a*f(a)-a²f'(a)-f²(a)/f'(a)=2*A Using Matlab (I tried to solve the ED myself but i couldn't do it) the general solution for f is: f(a)=±C/2-C²/(8*A)*a Where all solutions are lines that wrap around the (particular solution) curve: f(a)=A/(2a) Therefore, the only functions that satisfy that the area of the triangle formed by the x and y axes and the tangent line to any point of it is constant are the straight lines and the one with the form y=C/x, where C is a constant that generates an area of 2*|C|.

@ВасилийТёркин-к8х 2 года назад

How come A/(2a) doesn't satisfy general solution at any C?

@jpgm11 2 года назад

@@ВасилийТёркин-к8х f(a)=A/(2a) is the singular solution of the differential equation, so it can not be obtained from the general solution. To find this solution you have to find a C(x) value so that ∂[F(C,x,y)]/∂C=0, where F(C,x,y)=0 is the general solution. In this case we have: F(C,a,f)=f+C²/(8*A)*a±C/2=0 So: ∂F/∂C=C/(4*A)*a±1/2=0 C/(4*A)*a=±1/2 C(a)=±2*A/a And replacing in the general solution we obtain the singular solution: f(a)=±C/2-C²/(8*A)*a f(a)=(2*A/a)/2-(2*A/a)²/(8*A)*a f(a)=A/a-A/(2*a) f(a)=A/(2*a)

@ArjunKumar-king06 2 года назад

I totally love your videos.....They are the best and they are vivid to understand!

@jacklardner8229 2 года назад

One of my favorite questions from Calc 1

@subi-dev 2 года назад

The rabbit hole actually goes WAY deeper with this if you use powers other than -1. If you do use calculus, you can make a graph (in Desmos) where *x* represents what *a* is in this video and *y* represents the area under the tangent line according to *a.* If the power you use is *n,* then the resulting graph will look like some coefficient times *x^(n+1).* If that isn’t insane I don’t know what is. I haven’t figured out what the coefficients are though, and I have no clue about a pattern for functions other than *x^n.* Thanks for reading.

@tanavposwal 2 года назад

Keep posting these special obsevation it helps a lots in exams

@pingpong3311 2 года назад

1/x² would be 9/4a. Deriving the function would give you -2/x³. Your coordinate would be (a, 1/a²) Your slope would be -2/a³ Using the formula y-y1=m(x-x1), you would get y-1/a²=-2/a³(x - a). Distributing will give you y-1/a²=-2x/a³+2/a². Isolating y, you would get y=-2x/a³+2/a²+1/a², or y=-2x/a³+3/a², the equation in y=mx+b form. The y-intercept (also the height) is 3/a². Now we must find the base. If you set -2x/a³+3/a²=0, isolating for x, you will get x=3a/2. This is also the base. Last but not least, we have to now use A=bh/2. Plugging in the values, you will get A=(3a/2)(3/a²)/2. A=(9/2a)/2 Multiplying top and bottom by 2a will get you 9/4a. You're welcome.

@yagof6365 2 года назад

Great explanation, professor. Thanks a lot!!!

@chuckaviator6423 2 года назад

For the area under the line tangent to f(x)=1/x², one can apply the same method to get the answer. In this case, the area will be 2.25/a where a is the x coordinate of the point where the line is tangent to the function. I'll try to find an answer for f(x)=1/x^n as well and update. Update: For f(x)=x^-n the area under tangent can be found with the formula ((n+1)/(2a^n))×((a/n)+a) here a is the x coordinate of the point where the tangent meets I tried this out on desmos and it works nicely

@75blackviking 2 года назад

I love this guy's channel. He is so good at his profession.

@blackpenredpen 2 года назад

Thanks!

@robertprobst3836 2 года назад

This problem was part of the section with random standalone problems in one of the 12th grade mathematics Abitur exams of East Germany in the 1960s.

@prgrmr8666 2 года назад

I got this question last semester I solved it this way: 1) you name the point where the tangent line crosses the x,y axis, lets call them a,b 2) suppose the tangent point at (x1, y1) Now we have three points (x1,y1) (0,a) (b,0) Now we calculate the slope for the line between the first two points, and the slope between the first and third point using delta(y)/delta(x) Now guess what, they're equal to each other and they= the derivative at x1 Now you have three equations to solve to get the value of a, b which means you have the dimensions of the triangle Sorry if anything wasn't clear

@davidgillies620 2 года назад

If f is a function that satisfies the necessary concavity/smoothness requirements etc. then A(f), the area bounded by the tangent line through the point {a, f(a)} and the axes, is a functional -(f(a) - a f'(a))^2/(2 f'(a)). That can only be a constant for f' ~ f^2.

@tenthmascot 2 года назад

There's a nice geometrical way to show that the area of the triangle doesn't depend on the specific tangent line chosen (once you know that, you can pick your favorite tangent line and see that the area is 2). Consider the tangent lines at (a, 1/a) and (b, 1/b); we'll show the triangles they form with the axes have the same area. Consider the transformation obtained by scaling the x-coordinate by b/a and the y-coordinate by a/b. This transformation maps (a, 1/a) to (b, 1/b), and multiplies areas by a factor of b/a * a/b = 1; i.e. it keeps areas invariant. Now, consider the equation y = 1/x; under this transformation, it becomes y/(a/b) = 1/[x/(b/a)], or y/(a/b) = (b/a)/x, or just y = 1/x again; that is, the graph of y = 1/x is invariant under this transform. Summarizing, this transform fixes areas, fixes the graph of y = 1/x, and maps (a, 1/a) to (b, 1/b); thus, it maps the tangent line to (a, 1/a) to the tangent line at (b, 1/b). It's not difficult to see that the x-axis and y-axis are preserved as well, so this area-preserving transform maps the triangle for (a, 1/a) to the triangle for (b, 1/b). Thus these two triangles have the same area, as wanted.

@vijaykulhari_IITB 2 года назад

Sir you teach enjoying math🥳🥳

@Jha-s-kitchen 2 года назад

Learning school maths as RU-vid videos and facts is really awesome... I learnt Basic calculus from 3b1b and some adv. calculus and integrations from blackpenredpen, just calculus or just, by him... you are great.

@martin.thogersen 2 года назад

@blackpenredpen here is a challenge! Imagine "ANY shape" S. Scale it in the x-direction by a, and y-direction by 1/a to get shape S(a). Then scale S(a) by the _largest_ factor in both x- and y-direction, so it can be inscribed in the domain D bounded by f(x)=1/x, y=0, x=0. Call it S_max(a). Is Area(S_max(a))=const? It holds both for the rectangle, and your (tangent) triangle. It also holds for a flipped triangle (half rectangle). And a rectangle with a hole in it. And a circle/oval. The shape defined by the letters BPRP? What are the minimal shape requirements for this to be true? It can also work for infinite shapes (with finite area): S=D works. Are there other examples?

@michaelmounts1269 2 года назад

really liked this…very elegant. you flashed 3 diagrams showing differing tangent points…kind of wished you displayed for s few more seconds

@TheCrower77 2 года назад

im amazed at his ability to seamlessly switch between markers

@gregthomson1064 2 года назад

It’s pretty impressive how flawlessly he switches marker colors.

@michellauzon4640 2 года назад

Let (0 , u) , (v , 0) and (a , 1/a) be the points on the axes and the point on the curve. We have au + v/a = uv (from the line equation) and a = sqrt(v/u) (from the slope equation) So 2sqrt(uv) = uv => uv = 4.

@DarkBoo007 2 года назад

That pause at the y-intercept, I know you wanted to say, "Its right there" so badly lol

@Jkauppa 2 года назад

hey, the (line) integral of the function line int[0,1] range of f(x)=b*sqrt(1-x^2), yes, its a circle or ellipse, b=1 or b1

@hvok99 2 года назад

This is so cool! I love how clean this problem is. Thanks for sharing

@blurryface602 7 месяцев назад

An interesting property of hyperbolas is that area of the triangle formed by tangent and asymptotes of a general hyperbola [(x²/a²)-(y²/b²)=1] is equal to 2ab Here, if we have a=b, we get a rectangular hyperbola A special case of the rectangular hyperbola is of the form (xy=c²) Our equation, y=1/x is of this form. Thus the area of that triangle becomes 2c², which when c=1, is 2. Just another approach though...

@AJ-et3vf 2 года назад

Awesome video! Thank you!

@koeielul112 2 года назад

Things drawing attention: 1) Plush Pokemon icon 2) board marker juggling 3) Actual math stuff :D

@blackpenredpen 2 года назад

😆

@QuantumOverlord 2 года назад

Had a quick go at proving this before watching the video. This is actually a great question that requires a fairly rudimentary knowledge of calculus but will be quite challenging at the same time. Would be good as a stretch exercise.

@zhelyo_physics 2 года назад

This is really interesting! 😀

@blackpenredpen 2 года назад

😆 glad you like it!

@marsvandeplaneet7786 2 года назад

I always did it aproach as rectangle from (0,0) tot the (x,y) of the dot, with surface 1

@durianboi4565 2 года назад

Damn, the y-intercept got him really good lmao. Video was both interesting and entertaining, Thank you!!

@blackpenredpen 2 года назад

hahaha

@sergeygaevoy6422 11 месяцев назад

y = -x/a^2 + 2/a y + x / a^2 = 2/a Multiply by a/2 = 1 / (2 / a) y / (2 / a) + x / (2 * a) = 1 so we have interceptions (2*a;0) (0;2/a) It is like an ellipse.

@JohnVKaravitis 2 года назад

MIND BLOWN!

@erdo4321 2 года назад

best math channel

@guyhoghton399 2 года назад

Just for the sake of variety, here is a method that doesn't use calculus. We take the general straight line function y = mx + c and equate it to y = 1/x for its points of intersection with the curve: 1/x = mx + c ⇒ mx² + cx - 1 = 0 ... ① Now isolate the tangent lines as the ones where both points of intersection coincide, i.e. where the discriminant of the quadratic equation ① is zero: c² + 4m = 0 ⇒ m = -c²/4 Hence the tangents to the curve have the form y = c - c²x/4 ... ② The y-intercept is c, and the x-intercept is given by: c - c²x/4 = 0 ⇒ x = 4/c The area of the triangle formed by any such tangent line with the axes is half the product of the intercepts: A = ½.c.4/c = 2 as required, independent of c. If preferred we can integrate ② between 0 and 4/c to get the area: A = ∫₀ ⁴/ᶜ (c - c²x/4)dx = [cx - ⅛c²x²]₀ ⁴/ᶜ = 4 - (c²/8)(16/c²) - 0 = 2.

@matthewgiannotti3355 2 года назад

The triangle below the x-axis for the tangent line to the function y=ln(x) is another function that has this property with (a,f(a)) as intercepts

@TaleshicMatera 2 года назад

I've seen solutions of y = Ax+B (cheeky) and y = 1/x (people proving that it's the "only" solution), but it also works for y = A/x, the area is just 2A---which is still a constant (the question was if there were any other functions with 'similar' properties not just Area = 2).

@anthonycannet1305 2 года назад

Pretty sure it only works for 1/x because working backwards locks us in to that formula. If the area = some constant C, in this case 2. Working backwards we get that the area of a triangle would work out to be 1/2(CA)(C/A) the A’s cancel to get 1/2CC as the area. But we also know the area is C so C=1/2(C^2). That’s 2C=C^2 or C=2. So an area of 2 is the only constant we can get this way. Meaning a Y intercept of 2/A and an x intercept of 2A are the only way to fit the property. As we plot the point at X=A of the line with Y intercept 2/A and X intercept of 2A for every value of A it will only trace out 1/X

@E-Scholar_1 2 года назад

Area of this triangle is 2 "ahaa".

@blackpenredpen 2 года назад

😆

@Miksarxe 2 года назад

it seems that if you have an inverse power function f(x)=1/x^n the equation for the tangent area becomes A=1/2 *((n+1)^2)/(na^(n-1)). In the case of n=1, a is irrelevant but only in that case

@anshumanagrawal346 2 года назад

Another way to see it: Because the equation y=1/x is symmetric in x and y (which also means its graph is symmetric about the line y=x), the tangents at the points (a, 1/a) and (1/a, a) would enclose the same area with the axes, so it can't be proportional to a

@Robert_H. 2 года назад

The function f(x) has at the point (a|f(a)) the tangent t(x) = f'(a) * (x-a) + f(a). The zero point of the tangent is xN = -f(a)/f'(a) + a. The area that this encloses with the axes is calculated here as follows: A(a) = (t(0) * xN)/2 = (-f'(a) * a + f(a)) * (a * f'(a) - f(a)) / (2 * f'(a)) = (-a^2 * f'(a)^2 + 2 * a * f'(a) * f(a) - f(a)^2) / (2 * f'(a)) = - (a * f'(a) - f(a))^2 / (2 * f'(a)) We now want the area A(a) to be constant regardless of the choice of a. Finally, using the notation y = f(x), we can formulate the following differential equation: (x * y' - y)^2 + 2 * A * y' = 0, where A is a real number corresponding to the desired constant area. If one chooses the approach y = c/x^b, then follows: c^2 * (b+1)^2 / x^(2*b) - 2 * A * c * b/x^(b+1) = 0 Obviously, 2*b = b + 1 must hold, so b = 1. Thus it follows: c = A/2. If you choose a polynomial of the nth degree as approach, you can easily show that they cannot satisfy the differential equation (with the exception of the trivial linear function). For a Taylor series (infinitely long polynomial), you find conditions for coefficients, but these ultimately lead only to the trivial solution. Means: f(x) = A/(2*x) is the only function which solves the differential equation. Therefore, there is no other function with this property.

@davidseed2939 2 года назад

Solve without calculus. Coordinates of tangent point ( a,1a) Area of rectangle from thst point to origin =1. Cut the rectangle diagonally down from yaxis to xaxis. Slide one copy up to sitting on top of rectamgle and one copy to the right of the rectzngle. The two small triangles each with area 1/2 b=a, h=1/a. NOTE.. incomplete see discussion

@adityansingla5656 2 года назад

But how can you prove that the copies are same as the rectangles obtained by tangent?

@bobtivnan 2 года назад

You are assuming that the other two rectangles form a tangent to the curve.

@davidseed2939 2 года назад

@@bobtivnan not two rectangles but two triangles and obviously they are similar since they are right angle triangles with a hypotenuse of the same slope. what is particular about this equation xy=1 or i think any equation xy=c, is that the two similar triangles are equal. I think that can be proved by the fact that the equation is symmetric in x and y, although i have not done that.

@bobtivnan 2 года назад

Yes, I meant triangles. They are obviously congruent. My point is that you have made an assumption that a tangent segment is formed. It may be possible to show that with a proof by contradiction, but it's a hole in your reasoning as it stands.

@davidseed2939 2 года назад

@@bobtivnan |\ | \ ~~. |__|_\ in that image above… the period is the tangent point. The origin at the bottom left the rectangle between them a single tangent lines runs between the intercepts on the axes. formed from \\\ for any curve the triangles must be similar, because the corresponding angles are equal. but in the case y=1/x, the two triangles are equal. and equal to half the rectangle. in another case, eg y=1/x^2 then the two triangles would be similar but not equal. For that assertion to be a proof, I need to show that the slope of the tangent to the function, is the slope of the diagonal of the rectangle. I hope to do that by virtue of the xy symmetry of the curve. But I haven’t done that. does it work for (x-r)^2 +(y-r)^2 =r^2 ?

@zach.a.fields Год назад

If my math is right, it turns out a general formula for area under the tangent line of 1/(x^n) in the first quadrant is: (((n+1)^2)(2n-1))/(2n(a^(n-1))) At n=1, like in this video, it simplifies down to 2/(a^(1-1)), which is obviously just 2. Cool little result!

@andrewporter1868 2 года назад

It would seem there may be a way to generally compute closed forms of \int_a^b f(t) dt using f'(x) by recursively constructing triangles and subtracting from the sum of the previous and current triangles' areas the conjunction their areas by computing an explicit formula. That would be really convenient.

@Milkyway_Squid 2 года назад

So someone else showed that this property doesn't work for x^-n n!=1, however this trivially works for any linear equation that doesn't go through (0,0). I'd be wondering if this works for y = a/(x-b)+c or even a/(x-b) + cx + d. However, this probably is limited to certain values of those constants. Take for example 1/(x+1)+1. At x=0 y = 2 and the area is 2. As x tends to infinity, the slope tends to 0 meaning the y intercept of the tangent approaches 1 as the x intercept approaches infinity.

@technowey 2 года назад

That is super cool! Thank you for making videos!

@blackpenredpen 2 года назад

😃

@yolanda6392 2 года назад

4:10 When you have too many tabs open so your computer starts lagging

@dqrksun 2 года назад

It can be simplified to a differential equation 2xy*y'-y^2-(y')^2*x^2=2Ny' where N is the area you want. If you plug in N as 4, 1/x satisfy the equation. So if you want a function like 1/x but the area is 3 just plug in N as 3 and solve the equation. The DE is reallyy hard tho

@マサフミ-g4k 2 года назад

Since I lived in Singapore I like your singaporish English (don’t know where you live but you sound like a Singaporean

@popcorn101cheese5 2 года назад

OK THIS VID MAKES Me enjoy math

@zildijannorbs5889 2 года назад

A friend of mine gave me an elegant problem after I showed him this one. A parabola is given by the equation: y = x^2 - 1 We’re considering all chords of this parabola such that they pass through the origin. Find the curve which represents the set of midpoints of those chords.

@attoparsec5206 2 года назад

nice problem. the answer is another parabola, y = 2*x^2

@zildijannorbs5889 2 года назад

@@attoparsec5206 exactly. I found this problem super elegant! Also tried to iterate it a bunch of times and ended up with somewhat patternish parabola equations. Failed to generalize though.

@mohammedbasharat9584 8 месяцев назад

Nice problem. Have you got an elegant proof? I tried sledgehammer. Started with a point (a, a^2-1) used y=mx to find equation of line and then been trying to find where this line intersects the curve again...will then find midpoints. Not sure it will work either

@gocomputing8529 2 года назад

Very interesting video. I have a question. Maybe it's a little late, but could you or someone probe using differential equations that C/x is the only function that achieves this property? I tried but I ended up with a non linear ODE (ordinary differential equation), and I don't know how to solve it. The equation is: 2A*y' = -(y - x*y')^2 If you substitute 1/x you can see it is a solution, but I would like to solve the equation with no prior information. Thank you very much.

@shortnr Год назад

I'm a year late, but I thought I'd mention that you had some serious Michael Penn energy when you fumbled your words and started your thought over again at about 4:15 :)

@HanginInSF 2 года назад

Dude is explaining calculus in his second language with perfect grammar. Also he's only using one hand to do it.

@lucasfabre2206 2 года назад

If f(x)=1/x^n, we get that the area of the triangle is equal to ((n+1)^2)/(2na^(n-1))

@gaeb-hd4lf 2 года назад

Interesting, does this have any engineering applications?

@ЛидийКлещельский-ь3х 2 года назад

Стандартная задача для 11-го класса. Уравнение касательной : y=F’(xo)*(x-xo)+F(xo) .для заданной функции - F(x)=1/x получаем : y(x)=-(1/xo^2)*(x-xo)+1/x0. Координаты пересечения этой прямой с осями координат : y1=y(0)=2/xo и x1=2*xo . S=0,5*y1*x1=2. И всё!! С уважением, Лидий.

@eustacenjeru7225 2 года назад

I love your teaching skills

@davidblauyoutube 2 года назад

The curve y = - cx + sqrt(2c A) has this property, with area A. Of course this is trivial, since the curve is a straight line and its tangent at any point is itself.