The diffusion equation | Week 12 | MIT 18.S191 Fall 2020 | Grant Sanderson

Поделиться
HTML-код
  • Опубликовано: 7 ноя 2024

Комментарии • 204

  • @bbugl
    @bbugl 4 года назад +479

    i see grant in the thumbnail, i watch. simple man and such...

    • @skaterope
      @skaterope 4 года назад

      yeahhh ! haha

    • @pafloxyq
      @pafloxyq 4 года назад +2

      we all do.!!

    • @jesselybianto8650
      @jesselybianto8650 4 года назад +3

      Likewise. He has that soothing voice.

    • @nuzlock4481
      @nuzlock4481 3 года назад +2

      @@jesselybianto8650 kinda like math asmr

    • @javiercmh
      @javiercmh 3 года назад +1

      Yes, no idea about diffusion, just wanted to watch him explaining it

  • @tangc
    @tangc 4 года назад +279

    fun fact: Grant’s voice is more famous than his face

    • @vikranttyagiRN
      @vikranttyagiRN 3 года назад +9

      I disagree. Everyone recognizes that plushy pie expressive face

    • @greenstatic9294
      @greenstatic9294 3 года назад +26

      I had never seen his face before, I literally kept thinking "This guy sounds like 3 blue 1 brown" lmao

    • @tangc
      @tangc 3 года назад

      @@vikranttyagiRN lol

    • @ne0ck237
      @ne0ck237 3 года назад

      @@greenstatic9294 +++++

    • @jelco
      @jelco 3 года назад

      Khan academy as well?

  • @Zacks.C-land
    @Zacks.C-land 4 года назад +420

    Dude, you’re the voice of Bob Ross for math and science.
    We’ve got a happy little “differential equation” over here 🎨 and some excited molecules over here 🖼.

    • @JosephCatrambone
      @JosephCatrambone 4 года назад +8

      Check out his channel, 3Blue1Brown. Lots more of this.

    • @RAyLV17
      @RAyLV17 3 года назад +2

      There's another channel who does Bob Ross styled math explanations. Check out Tibees

    • @bamischijf_2757
      @bamischijf_2757 3 года назад

      And a sad calculus student over there.

  • @brendanc7278
    @brendanc7278 3 года назад +29

    Grant you NEED to do a series on PDEs. You are the best teacher I have ever encountered.

  • @ggaribarba
    @ggaribarba 3 года назад +10

    "by the time we actually start doing something meaningful and physical with it we just turn it right back into a discrete one". Very insightful, thanks!

  • @majorfallacy5926
    @majorfallacy5926 3 года назад +58

    Me: procrastinates studying heat and mass transfer all day
    also me: oh look a grant sanderson video *click*

  • @Tobaman111
    @Tobaman111 4 года назад +56

    Excellent. I feel less dense already.

  • @subhasish-m
    @subhasish-m 4 года назад +38

    The intuition and derivation for the heat equation was superb! It felt very natural

  • @JonasGranel
    @JonasGranel 3 года назад +25

    Waaaait a minute! Isn't that the 3blue1brown guy?

  • @MichaelG-gg2bz
    @MichaelG-gg2bz 3 года назад +2

    Clicked for Grant, was surprised by intuitive explanation of stochastic calculus...this man is the gift that keeps on giving.

  • @stefanobutelli3588
    @stefanobutelli3588 4 года назад +10

    Very nice. As always you could be the greatest math divulgator of our time. This remembers me of my Stochastic Processes exam. From a symmetrical random walk (in 1D means same probability to go left or right) one can derive the diffusion equation, and from the asymmetrical version you can derive the Fokker-Planck equation, aka the diffusion-with-wind-equation :)

  • @eason3892
    @eason3892 4 года назад +3

    This is the most intuitive explanation of the diffusion equation. I finally get a rough idea of it this time. Thanks, Grant

  • @Quotenbrtchen
    @Quotenbrtchen 4 года назад +9

    Once you've seen this derivation, you'll always remember the unit of the diffusivity coefficient. Very cool stuff!

  • @umedina98
    @umedina98 3 года назад +1

    Man your explanations cant stop surprising me! I can keep thanking you for explaining in such an intuitive way. You feel the passion and the love for what you are doing.

  • @dfunited1
    @dfunited1 4 года назад +5

    Oh boy did I not continue deep enough into Calculus. I learned a lot, and will save this to my favorites for rewatching. Give this man some bongos.

  • @OurNewestMember
    @OurNewestMember 3 года назад

    Probably the best 20 minutes I've spent on RUclips. I could just rant on the insight and intellectual value. Cheers! 👏👏👏

  • @bahamut149
    @bahamut149 4 года назад +74

    Me as a high school student: I can understand this equation.
    Me after university: Let me open my calculator app and check again if 6+7=13.

    • @skyli3945
      @skyli3945 3 года назад

      damn multi-variable cal in high school lol

  • @emperorpingusmathchannel5365
    @emperorpingusmathchannel5365 4 года назад +175

    I think Grant is the feynman of 21st century when it comes to teaching.

  • @micahcantor7356
    @micahcantor7356 4 года назад +5

    Wow, that 3D graph of the surface was incredibly well visualized.

    • @gileee
      @gileee 4 года назад

      That's what Grant, the dude in the video, does. 3Blue1Brown is his RUclips channel. One of the best math education/visualization channels on this whole site.

    • @micahcantor7356
      @micahcantor7356 3 года назад +1

      @@gileee Oh I know. I've played around with manim (the software library he uses to make the animations) myself, but Grant is still by far the king of making visualizations like these.

  • @laurentiustoleriu8110
    @laurentiustoleriu8110 4 года назад +13

    Very high quality material. Just a small remark from a physicist after listening the first minute: heat is the one that flows, not the temperature.

  • @xintongbian
    @xintongbian 4 года назад +6

    for the 1st time in my life i see the hope of understanding laplacian... it freaked me out every time when i try to understand graph convolution

  • @andrewlesak236
    @andrewlesak236 4 года назад +2

    loved watching the smooth transition from a simple idea to an equation

  • @hooeezit
    @hooeezit 4 года назад +1

    So beautifully and lucidly illustrated! I thank the RUclips algorithm for bringing me to this channel.

  • @marccox8977
    @marccox8977 4 года назад +7

    Thanks Grant ! One take away for me : Directly-Unsolvable PDE, Numerical simulations to the rescue !

  • @Noceo
    @Noceo 4 года назад +11

    This video just made me less dense. Loving it, thanks Grant!

  • @likithstochastic
    @likithstochastic Год назад

    15:00 - It's a nice way to look at the diffusion equation! The rate of change of the variable (partial derivative) is directly proportional to the curvature of the curve (as a function of the space), with Diffusion Coefficient being the proportionality constant. The parts of the curve where the magnitudes of the curvatures are high experience higher time rate of change. This explains why the curve smoothens out, or in other words 'straightens out' as time goes by.

  • @mingyang2128
    @mingyang2128 4 года назад +15

    So beautiful! We also use this equation in population genetics to study how allele frequency changes along with time.

    • @Tobaman111
      @Tobaman111 4 года назад +1

      I'm new to the field. Could you recommend a reference with implementation?

  • @msergejev
    @msergejev 3 года назад

    It is a revolutionary little equation. Basically, what it states is - the higher the inequality, the higher the incentive for change.

  • @DrKjoergoe
    @DrKjoergoe 3 года назад

    19:00 oh wow, this really opened my eyes regarding why exactly the mean value formula for harmonic functions works!

  • @DuniC0
    @DuniC0 3 года назад

    The "statistical noise" of 4:52 is not such, it isn't because the "random statistics" ( 4:38 ).
    It's actually because the implementation of the random walk explained at 3:52 :
    50% hop to the left, 50% hop to the right
    The resulting curve would be smoother:
    - if we interpolate the values between steps
    - if we add a third option: 33% hop to the left, 33% stay, 33% hop to the right
    - Even better if it had 3 options with a discrete normal distribution, ie: 25% hop to the left, 50% stay, 25% hop to the right
    - The best would be a continuous normal distribution

  • @imdadood5705
    @imdadood5705 3 года назад +7

    If I had his voice, I’d never stop talking! 😂

  • @miro.s
    @miro.s 3 года назад

    Inspiring video. It made me to think about generalized Pythagorean theorem coming from duality in linear spaces. The divergence is another result of the duality that is coming from linearity it is keeping invariant and doesn't change with unitary transformations.

  • @mengyuwang5159
    @mengyuwang5159 3 года назад

    Gorgeous way of teaching diffusion equation!

  • @saxy1player
    @saxy1player 4 года назад

    I like the approach (thinking about random walk) and the clarity of the explanation :)

  • @whatitmeans
    @whatitmeans 3 года назад

    When you start talking about the brownian motion and then later appears de squared of the delta_x I was waiting for you to start talking about how this thing are managed to be constant allowing the solution truly exist, introducing in this way you viewers into Ito's Calculus.... hope you can extend the video with this modern view of Brownian motions analysis. Thanks you beforehand!

  • @isbestlizard
    @isbestlizard 4 года назад +24

    HEY I KNOW THIS VOICE :O

  • @marouaniAymen
    @marouaniAymen 2 года назад

    I liked this video, but why the Schrodinger Equation is a diffusion equation ? Perhaps the Wave Function is diffusing in space, any idea ?

  • @oraz.
    @oraz. 4 года назад +3

    Frikkin wizard-man, this guy!

  • @muknunjoon
    @muknunjoon 7 месяцев назад

    How does this man teach everything I learn?

  • @1q1q1q1q1q1q1qw
    @1q1q1q1q1q1q1qw 3 года назад

    i really like your style and appreciate it that you explain julia

  • @samlopezruiz
    @samlopezruiz 2 года назад

    Brilliant animations as always!

  • @rif6876
    @rif6876 3 года назад +1

    And I think that if you give the time dependent side an imaginary number, you get the schrodinger equation from quantum mechanics.

  • @PapaFlammy69
    @PapaFlammy69 4 года назад +297

    11111

  • @tempdeltavalue
    @tempdeltavalue 2 года назад +1

    A bit unclear why we have dX^2 in denominator when we take difference of differences 😢

  • @grainfrizz
    @grainfrizz 3 года назад

    It already clicked when you increased to dots to 10,000. That's why you're a great teacher, Grant.
    P.s. missed 3b1b.

  • @stefano8936
    @stefano8936 3 года назад

    The temperature do not just GENERALLY flow from the warm part to the cold part. It's the first law of thermodynamics.

  • @BTElectric
    @BTElectric 3 года назад +1

    Ohhhh nice. I'm in a semiconductor course and was wondering how ficks law came to be.

  • @dream-sage1017
    @dream-sage1017 4 года назад

    It was great. Your videos are amazing. I hope you continue to upload these kind of videos.

  • @uzdik.student
    @uzdik.student 4 года назад +1

    Is it a Blue Yeti microphone?

  • @tranly6883
    @tranly6883 3 года назад

    first time here and worth it, i see grant's handwriting

  • @farrankhawaja9856
    @farrankhawaja9856 4 года назад +2

    Ahhhh, so this is the '1brown'. Love your videos so much!

  • @GalinaLipunova
    @GalinaLipunova 4 месяца назад

    20:30 In some cases you can solve this equation exactly (Green functions, etc.). So you can kind of check your computer code against these solutions

  • @smal7812
    @smal7812 3 года назад +1

    No idea how this ended up in my Recommended at 01:15 am but I did like it.

  • @leonardosandovalpabon6561
    @leonardosandovalpabon6561 3 года назад

    Hi! Why did you define the number of particles in the boxes as densities rather than concentrations? Are you assuming that the particles occupy all the volume of the boxes?

  • @GraphiSx
    @GraphiSx 3 года назад

    Finite differences, it makes more to use finite volume schemes, specificly when D is not a constant

  • @BryanLeeWilliams
    @BryanLeeWilliams 4 года назад +4

    Wow, RUclips has never shown me these. And I never noticed them on Twitter I guess.

  • @johndwolynetz6495
    @johndwolynetz6495 4 года назад +13

    this makes a math nerd happy 😭

  • @trivialstuff2384
    @trivialstuff2384 4 года назад +5

    I'll be Happy If He writes a book on Mathematics.

  • @bitansarkar6463
    @bitansarkar6463 3 года назад +5

    when will you come back to 3blue1brown :(

    • @HaydenDoingThings
      @HaydenDoingThings 3 года назад

      He's producing lectures for MIT... obviously that's much more prestigious than a RUclips channel.

  • @andrewkreisher689
    @andrewkreisher689 3 года назад

    I took this class last semester wish i would have waited for this one!!!

  • @scichord
    @scichord 3 года назад

    Anyone have any good resources (books/simple papers) for doing this kind of thing on a finite grid? I've been trying this with random walks on 8-neighbors and it keeps drifting and settling in the center and I'm not sure why that is.

  • @gustavoexel5569
    @gustavoexel5569 3 года назад +1

    It'd be nice to see some Finite Volumes Method solving the diffusion problem...

  • @OldManSilencer
    @OldManSilencer 3 года назад +1

    your voice sounds like 3 blue 1 brown. good lecture but was surprised to hear that exact voice cadence.

  • @arnavrawat9864
    @arnavrawat9864 3 года назад

    Is there a course for differential equations

  • @UmarUmar-bg1fj
    @UmarUmar-bg1fj 2 года назад

    Excellent channel

  • @lemsolaris67
    @lemsolaris67 3 года назад

    Great ! Makes me feel mathematical young, but I’m not young I did these exams 30 years ago, I wish I had this fun at the time

  • @int16_t
    @int16_t 3 года назад

    Everyone is commenting about him, but no one is talking about the topic. Probably because it was explained clearly.

  • @ezioauditore4109
    @ezioauditore4109 7 месяцев назад

    why did the uppercase D disappear

  • @Guytron95
    @Guytron95 3 года назад

    Bonus: the diffusion equation plays a central role in Grover's function inversion algorithm in quantum computing.

  • @quaidcarlobulloch9300
    @quaidcarlobulloch9300 4 года назад +1

    Brilliant content, thank you!

  • @robertgalindo5979
    @robertgalindo5979 3 года назад

    3blue1brown has a second channel? I had no idea.

  • @Gunth0r
    @Gunth0r 3 года назад +1

    you had me at "hot rod"

  • @apostolosvrontos5792
    @apostolosvrontos5792 4 года назад

    Very beautiful explanation video. Now you can blink

  • @sajidhaniff01
    @sajidhaniff01 2 года назад

    Awesome! Thanks Grant

  • @hamzamameche3893
    @hamzamameche3893 3 года назад

    Hi, I'd like to know how did you make the graph animation
    thanks

    • @nbme-answers
      @nbme-answers 3 года назад

      it's his own library called Manim ("math animations"); it is open source

  • @BRunoAWAY
    @BRunoAWAY 4 года назад

    I prefer using conservation of mass in integral form and the divergence theorem tô deduce the equation

  • @asadmamun5772
    @asadmamun5772 2 года назад

    Amazing explanation. Thanks a lot.

  • @asifequbal8390
    @asifequbal8390 3 года назад

    How can I make such animation?

  • @jewulo
    @jewulo 4 года назад

    Dear Grant,
    What software do you use to create the interactive graphical animations for your videos? Is this commercial software? Perhaps, did you write the software? If so are you able to opensource for those interested in the code? Thanks as always from a keen learner from your work.
    Regards,
    Joshua

    • @tarekacila9204
      @tarekacila9204 Год назад

      He uses a library called Manim in Python. As far as I know, he is the one who founded this library.

  • @yangwang3687
    @yangwang3687 10 месяцев назад

    You have good looking, good voice and good teaching!

  • @Abhishek-hy8xe
    @Abhishek-hy8xe 4 года назад +1

    Temperature does not flow. Its eat.

  • @RobertoMariani
    @RobertoMariani 3 года назад

    3 Blue 1 Brown is DIFFUSING!

  • @tariq3erwa
    @tariq3erwa 4 года назад

    2:54 if you are watching on phone and full screen and jiggle the phone a bit while looking at the dot It'll make a nice effect

  • @saiputcha1730
    @saiputcha1730 3 года назад

    What if the fraction of the density variation is not 1/2?

  • @rodrigoserafim8834
    @rodrigoserafim8834 3 года назад

    Rho, rho, rho your boat
    Gently down the slope
    Merrily, merrily, merrily, merrily
    Life is but a delta

  • @gilmartrevisan
    @gilmartrevisan 3 года назад

    Great class, professor.

  • @randomrandom4535
    @randomrandom4535 3 года назад

    Absolutely perfect.

  • @ilustrado7291
    @ilustrado7291 3 года назад

    This is Pewdiepie if Pewdiepie is a mathematician.

  • @Lyr00
    @Lyr00 3 года назад +2

    0:06 guess i now know why his channel is named "3blue1brown. and i got the same eyes as him, just the 3 blue is grey for me and i also have a smaller brown dot in my other eye

  • @arjayUU
    @arjayUU 4 года назад +1

    Somebody buy this man a nice pen and squared paper!

  • @RajanNarasimhan
    @RajanNarasimhan 3 года назад

    Incredible!

  • @jhanolaer8286
    @jhanolaer8286 2 года назад

    Sir,please teach us also about advection..please😥

  • @livedandletdie
    @livedandletdie 3 года назад

    So that's why cityfolks where I live can't understand the keep your distance signs. Because they're too dense. Which means they don't keep distance...

  • @iammanishverma
    @iammanishverma 3 года назад

    Very well explained !!

  • @SaveSoilSaveSoil
    @SaveSoilSaveSoil 2 года назад

    This was helpful! Thank you!

  • @TheKaylotta
    @TheKaylotta 4 года назад

    amazing explanation !

  • @maharajxx
    @maharajxx 4 года назад

    Why would (delta x)^2/(delta t) be constant? (delta x)/(delta t) is molecules velocity and may be considered a constant value. But what about another delta x left?

    • @3blue1brown
      @3blue1brown 4 года назад +5

      Great question! If each particle is moving according to a random walk, then after many steps the distance traveled will not be proportional to the time. However, the *square* of the distance traveled will, in expectation, grow linearly with time. It's a fun little exercise to work out.

    • @maharajxx
      @maharajxx 4 года назад

      @@3blue1brown Please correct me if I'm wrong. Square of drift is proportional to time in the case of, strictly speaking, delta t (hence delta x) -> Inf. Here, though, we are dealing with small increments. Notwithstanding all this minutia, I am a big fun of your brilliant channel. Keep up your excellent work!

    • @TJStellmach
      @TJStellmach 4 года назад +1

      @@maharajxx also consider that as dx->0, the number of cells in a given volume of space goes to ∞. It's the granularity of x and t, not their magnitudes, that makes the random walk statistically a normal distribution, which has the squared relationship in the expectation that's at issue here.

    • @piotr.ziolo.
      @piotr.ziolo. 3 года назад

      @@3blue1brown Is the constant of proportion just an arbitrary choice of the person creating the mathematical model or is there any connection to the real material that is diffusing? It seems like this is the only link missing between mathematics and physics in your presentation. I wonder if it can be filled in a mathematically rigorous way.

  • @Pedritox0953
    @Pedritox0953 4 года назад

    AWESOME EXPLANATION!!!

  • @aviavi4684
    @aviavi4684 3 года назад

    3blue1Brown ?

  • @osho_magic
    @osho_magic 2 года назад

    hello,I am very happy that I found you, I am doing distant course in DSML in collaboration with MIT , I wish if i too had a teacher like you , any suggestion for me , i am very much interested in deep learning and ai , and i know your lectures will be very useful for me and thanks for making them available for free ....