Home Creators Posts Import Register Favorites Logout
haven't archived this post yet. have a subscription? use the importer!

Content

This is the follow-on to the Newton's fractal video. I tried to frame it in such a way that it doesn't necessarily require watching the first as a prerequisite, so perhaps I should simply say here's the next lesson. In either case, let me know if you catch any errors or have any feedback!

Files

Previews only

Where Newton meets Mandelbrot (Holomorphic dynamics)

How the right question about Newton's method results in a Mandelbrot set. Video on Newton's fractal: https://youtu.be/T_S2j5GaLRQ Special thanks: https://3b1b.co/lessons/newtons-fractal#thanks Extra special thanks to Sergey Shemyakov, of Aix-Marseille University, for helpful conversations and for introducing me to this phenomenon. ------------------ Introduction to Fatou sets and Julia sets, including a discussion of Montel's theorem and its consequences: http://www.math.stonybrook.edu/~scott/Papers/India/Fatou-Julia.pdf Numberphile with Ben Sparks on the Mandelbrot set: https://youtu.be/FFftmWSzgmk Excellent article on Acko.net, from the basics of building up complex numbers to Julia sets. https://acko.net/blog/how-to-fold-a-julia-fractal/ Bit of a side note, but if you want an exceedingly beautiful rendering of the quaternion-version of Julia fractals, take a look at this Inigo Quilez video: https://www.youtube.com/watch?v=rQ2bnU4dkso I first saw Fatou's theorem in this article: https://projecteuclid.org/journals/communications-in-mathematical-physics/volume-91/issue-2/On-the-iteration-of-a-rational-function--computer-experiments/cmp/1103940533.pdf Moduli spaces of Newton maps: https://arxiv.org/pdf/1512.05098.pdf On Montel's theorem: https://people.ucsc.edu/~fmonard/Sp17_Math207/lecture11.pdf On Newton's Fractal: https://core.ac.uk/download/pdf/1633889.pdf ------------------ These animations are largely made using a custom python library, manim. See the FAQ comments here: https://www.3blue1brown.com/faq#manim https://github.com/3b1b/manim https://github.com/ManimCommunity/manim/ You can find code for specific videos and projects here: https://github.com/3b1b/videos/ Music by Vincent Rubinetti. https://www.vincentrubinetti.com/ Download the music on Bandcamp: https://vincerubinetti.bandcamp.com/album/the-music-of-3blue1brown Stream the music on Spotify: https://open.spotify.com/album/1dVyjwS8FBqXhRunaG5W5u ------------------ Timestamps: 0:00 - Intro 3:02 - Rational functions 4:15 - The Mandelbrot set 8:12 - Fixed points and stability 12:51 - Cycles 16:25 - Hidden Mandelbrot 21:17 - Fatou sets and Julia sets 26:24 - Final thoughts ------------------ 3blue1brown is a channel about animating math, in all senses of the word animate. And you know the drill with YouTube, if you want to stay posted on new videos, subscribe: http://3b1b.co/subscribe Various social media stuffs: Website: https://www.3blue1brown.com Twitter: https://twitter.com/3blue1brown Reddit: https://www.reddit.com/r/3blue1brown Instagram: https://www.instagram.com/3blue1brown_animations/ Patreon: https://patreon.com/3blue1brown Facebook: https://www.facebook.com/3blue1brown

Comments

Anonymous

Thanks for posting! Watching now. I noticed one graphical glitch at about 4:09, where the rational function expression flickers between the general formulation and z^2+c. Edit: there's also a single frame glitch around 25:54.

Anonymous

Fantastic, as always. Thanks. I think I found three minor typos: - At 3:20, there is a superfluous plus sign before the 2z^3 (rightmost expression). - At 12:54, exercise 1c, the "c" in "values c satisfying" should be typeset as $c$. - At 21:39, "contains 1 colors" -> "contains 1 color"

Jesse Thompson

11:15 Grant: "It *looks* like multiplication by 2i" Me: "It *looks* like a brilliant demonstration about how to further improve manem's interpolation: what if instead of linearly moving every point to its destination, one moves the points along a path that also interpolates the derivative? In this case, that would include the points not shrinking prior to expanding in the zoomed in view, but just inexorably expanding. :)

3blue1brown

That's definitely very doable, and a good suggestion. In fact, in past videos I've interpolated between the identity and z -> z^2 with z -> z^(1 + alpha), which I can throw in here.

Anonymous

4:08, where you talk about z^2+c, your graphic messes up here.

Christian Leichsenring

Could you put the video and article you reference in the description?

Lionel Pöffel

Very nice. One point: the video *mentions* that the principle "every point is either not on a boundary at all or on a boundary between all colors / asymptotic behaviors" does not explain Julia sets being fractal and not smooth. But then it goes into this different direction, showing that every disk around any point in the Julia set grows arbitrarily. No reason for the fractalness is actually given. Is the complicated Montel's theorem needed to explain this as well? Or would there be a simpler explanation.

Lionel Pöffel

Interesting fact about Julia sets that (if I remember correctly) isn't mentioned: the Mandelbrot set is a map of all Julia sets, where points inside the Mandelbrot set correspond to Julia sets whose "interior" (black region) is a solid, continuous region (or at least composed of some), whereas points outside corresponds to Julia sets whose interior is a cantor dust (totally unconnected set).

Anonymous

Man that sent me to a very deep rabbit hole. Amazing job as usual! Minor error, at 16:30, exercise 2 has the numerator as z^3 -2z + z where as it should be z^3 -2z + 2

Anonymous

There is a typo at 16:27. Numerator should have z^3-2z+__2__. Love all the exercises btw. Great video!

Eric Severson

Beautiful as always, and very glad I was able to watch the followup immediately. I'm curious just how recent your discovery of all this material was. I know there was a bunch of points from the patreon comments on the Newton's Method video asking about Mandelbrot connections. Did you know how precise those connections were at the time people were asking that? From my end, it was immensely satisfying to have seen those questions along with the first video, and then to have this come along so soon after with such a gorgeous explanation.

RAD Donato

Thanks for another great video, Grant. I don't have near the math skills to critique our help out, so just thanks for great content.

C.J. Smith

7:12 OMG these visualizations are ridiculous (in the good way)

Anonymous

Hey Grant, I'm loving this mini-series! Around 3:40-3:50, the language of iterating rational functions is very central during the time that the boxed equations seem to be changing through various examples of rational functions. This is a little confusing in the sense that a verb you're using and the visual event on the screen are somewhat more incongruous than is usually the case. Thank you for this joyful visual experience of abstract mathematics! Daniel

Anonymous

At 4:09 the "z^2+c" switches back to the general expression for a couple of seconds. Not sure if that was intentional.

Andrew Alvarez

I loved this video!! I have a couple of nits I'd like to address, and it looks like these haven't been pointed out by anyone else yet: - 3:59 visual glitch where the size of the yellow box instantly changes - 4:22 the transition of the expression towards the left is a little choppy here - 10:00-11:55 it would probably be a good idea to connect the lessons here to the video you did a while back about stability (The other way to visualize derivatives) **NOTE: I just saw that you added the above-mentioned video to the "Essense of Calculus" series as Chapter 12, but in the title you call it "Chapter 2"! PLEASE FIX THIS!!** - 13:40 I think the word "minus" is confusing here - it sounds like you could be saying that it's set of solutions to [f(f(x)) - f(x)], though I know that's not your intention. Maybe you should either say "that are not" instead of "minus" here, or display the statement you're trying to make in set notation. - 18:50 "That one seed value is definitely going to find it". You're implying that 0 is always the centroid of the roots of the function z^2 + c, which is true, but some viewers may not know why so this might confuse them. You could probably also display the roots of z^2 + c, perhaps in red, alongside the yellow sequence points to mostly fix this. Or you could make proving this a mini-exercise, but you yourself mentioned feeling like you may be assigning too much homework here ;'). - 24:22 I'm just curious, what are the possible exceptions? Are they always randomly different depending on the Julia set, or are there particular points that are always of interest like the "point" at infinity or something?

Anonymous

Grant these two videos on fractals are amazing! I did my maths master thesis on holomorphic dynamics. I did it because I had questions about the mandelbrot set that popular expositions did not answer. Your video now answers so many of them in a clear and compelling way, without hiding the complexity away! I am amazed that you managed to digest the material to such an extent for the benefit of posterity. Kudos! I only wish you'd spend a little more time on the mandelbrot-like shapes in the parameter space for cubic polynomials, because I wished you'd answer some of the questions I didn't get to in my master's, like: - in the z^2+c mandelbrot set, the main cardiod represents attracting points, but in the cubic case, it represents cycles (i guess 2-cycles). What to the other parts of the set represent? - why are the mandelbrot sets in the cubic parameter space a bit "deformed"? what does it mean? - are there more mandelbrot sets around the same parameter space? how are they related? These are complex questions of course... related to renormalisation. Perhaps a topic for a new video? ;)

Rion Boom Crabhands Keon

Small suggestion: When you explain how the derivative is locally a linear transformation around 11:00 - 11:30, you might include a local coordinate system so that the rotation aspect is visible.

Anonymous

Awesome stuff, could you include a reference to Fatou's 1919 Theorem at 18:24 in the description?

3blue1brown

You're correct, this video only explains the case with 3+ stable attractors. And in fact, the case with 2 attractors can have a non-fractal boundary (e.g. Newton's method on quadratic, or f(z) = z^2). What's the right fix here, being extra clear about what is and isn't explained?

3blue1brown

It was actually one of the Patrons who pointed me to this Mandelbrot connection, which I had not known anything about. Benefits of the early releases!

3blue1brown

Thanks for the notes, good catches! Not sure how the calculus chapter was misnamed, but that's fixed now. For the z^2 + c example, my personal taste is that it's easier to see that f'(0) = 0 by display f'(z) = 2z, rather than showing the roots, but I suppose to each their own. To 24:22, it's not always the point at infinity, but one easy case is that where f(z) is a polynomial, one of those exceptional points definitely will be the point at infinity.

3blue1brown

You're not the first to suggest renormalization as a new video :) That could definitely be a fun one. After all, forget the Newton maps, just explaining why the Mandelbrot set itself has tiny (and deformed) mandelbrot sets inside it would be fun.

Anonymous

Nice Video! At 4:09 the z^2 + c / 1 disappears and then reappears a couple seconds later.

Don Sanderson

15:16:  Ooooo . . . you really kicked a hornet's nest THERE.  Doubling down, trolling for trolls, daring the teeming hordes of lightly-informed armchair philosophers to reflexively howl their pontifications about the improbability of the "probability being zero."  Gutsy move Mav.  Is your asbestos armor at hand?  Prepare for an onslaught of indignant rage:  Just sayin'.

Uri Agassi

It is mind boggling to realize where Mandelbrot Sets are hiding! At 16:34 - the hint regarding how to calculate the derivative disappears a bit too fast [at least for non-native english readers] to read.

Edith Dubiner

very nice and interesting - I like these two videos a lot! Maybe stress a bit more that the (black) Mandelbrot set appears only in the "left" fixed seed case (and not the fixed function).

Anonymous

You've just played with my math addiction and made me a Patreon T_T No, but more seriously, you're a true example to me, I often have tears while watching what you do.

James Matheson

That was a very interesting video, I reallly enjoyed it. There were two bits though that felt a little less well defined than usual when watching your finished videos. At 21:27 where you say "it (the madelbrot set) seems it has a more general relationship to parameter spaces" I would love to know more about what that relationship is (is it known, and just too complicated to explore in the video, or we genuinunly don't know why?). I would also be very curious to know how it change things in your function to paramter space example if the other two roots are elsewhere than 1 and -1. Again really awesome video. I love the way you explain things with visualisations, if that kind of thing had been around when I was in uni, I would probably still be in pure math (and not computer science ), as it is you are really drawing me back to my roots :-).

Sachin Shukla

Watching this video, and the last one, I’m reminded of your video on fractal dimension where you declared that fractals are about roughness, not self-similarity. This set of videos, I think, is a really good way to explain that statement. I think that message could be made stronger in this video.

PeetieGonzalez

2 major epiphanies from me upon seeing these videos. 1) The boundary is fractal in nature because there is no part of the boundary that isn't a boundary between ALL the roots of the polynomial. Already mind=blown. 2) The Mandelbrot set, if we think of it as the BOUNDARY between the area with finite cycles and the infinity trajectory, is the boundary between infinity and ALL number of finite cycles. And hence that boundary is also fractal in nature. This is the most informative and mindblowingly clear video on maths I have ever seen. I'm going to have to relegate your video on Euler's Identity (edit... actually the fourier transform, by extention) to my #3 spot (or #2 if I count this as one topic in two parts). So glad you're back.

PeetieGonzalez

Oh and seeing the familiar Mandelbrot beetle show up in such an unexpected place... I'm not easily spooked but that was almost a religious experience. Also obviously reinforces point 2 above. I'm just astounded that it shows up in such form.

Anonymous

-i is mis-labeled as "1i" on many (not all) of the axes.

Steven Siddals

Mind blowing! So many wonderful links. And I've wanted to see that side-by-side view of the Julia sets that arise from different points on the Mandelbrot set for a long time. Thank you Grant for another signature mix of beauty and depth! Just two thoughts for the final edit. Firstly I'd love to get any more insight into why it's a Mandelbrot of all things that shows up in that lambda fractal (especially as the title sets that up as the main teaser). And for my taste, poor Gaston Julia doesn't deserve to be ranked within the subset of mathematicians with similar unfortunate injuries...

3blue1brown

You can now play with that side-by-side on the website: https://www.3blue1brown.com/lessons/holomorphic-dynamics To your first point, I may do a follow-up to address that. To be honest, it's something I'm only recently familiar with as of this month, but the relevant keyword search is "renormalization", and it's similar in spirit to why the Mandelbrot set has smaller copies of itself as you zoom in. To your second point, you are right that was insensitive, my apologies.

Anonymous

I am having trouble reproducing your results. If I terminate the Newton solve after a large number of iterations (same approach to generating the Mandelbrot set), I can recreate your result. However, computing f'(z) = P(z)P''(z)/(P'(z))^2 always results in a zero when z = lambda/3. I get lambda/3 because z1 = 1 and z2 = -1 so (z1 + z2 + z3)/3 is z3/3.