This Differential Equation is Nuts

Поділитися
Вставка
  • Опубліковано 16 січ 2025

КОМЕНТАРІ • 92

  • @AlBoulley
    @AlBoulley 6 місяців тому +76

    "Life is like differentials and integration… you never know what the fuck you are going to get"
    -Flammy Gump

  • @zunaidparker
    @zunaidparker 6 місяців тому +34

    I think you're missing the cases where c=0 or c

  • @ralfbodemann1542
    @ralfbodemann1542 6 місяців тому +26

    Great. _You presented the solution for C > 0. But you also need to consider the cases C=0 und C< 0.
    BTW: I noticed LHS = d/dx(y^2/2) and RHS = d/dx(y'). If you tranformed the original equation that way, you could have immediately performed the first integration.

    • @elibrahimi1169
      @elibrahimi1169 6 місяців тому +1

      @@ralfbodemann1542 see that's what i am talking about, he probably just wanted to show another method

  • @liamschreibman8268
    @liamschreibman8268 6 місяців тому +48

    Hi, I'm a high school further maths student in London. I just want to say that I love your videos! Lots of people appreciate your work, myself included.

  • @dank.
    @dank. 6 місяців тому +5

    This differential equation reminded me a lot of your video on exactly solving for the period of an undamped pendulum. Mostly the start of that approach in recognizing the forms of derivatives, at least how I approached the problem (haven't watched the video yet).
    We can start by recognizing that the lhs looks pretty chain rule-y. We can express it as the derivative of (y^2)/2. The rhs is also just the derivative of y, so we can integrate both sides to get rid of the derivatives. This leaves us with (y^2)/2 = y'. We can move all the factors of y to one side. Doing this gives us (y')/(y^2) = 1/2. The lhs once again looks chain rule-y, and we can see that it can be rewritten as the derivative of -1/y. Making that substitution, we can once again integrate both sides to get -1/y = x/2 + c. Rearranging we get the final answer of y=(-2)/(x+c)
    Fun diffie-q!

  • @romanvolotov
    @romanvolotov 6 місяців тому +13

    my immediate thought was to recognise the original LHS as (½y²)'. then it's quite similar to the video. this thing often pops up in physics from what i can tell.
    thanks for the video

    • @brashmane2749
      @brashmane2749 6 місяців тому

      And it was very popular with several of my examinators... Especially with C belonging somewhere on Z...

    • @yung-wayliu2840
      @yung-wayliu2840 4 місяці тому

      Me too. Immediately bring the order down to first order, which is much more manageable.

  • @nonamehere9658
    @nonamehere9658 6 місяців тому +4

    At ~4:00 You've concluded something of the sorts: if A(x)*B(x)=0 for all x, then either A(x)=0 for all x or B(x)=0 for all x, which isn't necessarily true -- e.g. suppose a counterxample A(x) = [x==1], B(x) = [x=/=1] (this specific example isn't continuous, but there are other continuous A(x),B(x))
    I didn't follow very closely derivation for tan, but intuition tells me that y(x) can be a piecewise combination of constant function (on some subset of reals) + the tan from the second parth of the video on the other subset of the reals. Maybe can be even combined such that y(x) is smooth and y'(x) is also smooth.

    • @PixelSergey
      @PixelSergey 6 місяців тому

      I was wondering this as well! That step should not necessarily hold. I wonder how you would fix this step in the solution 😄

  • @charlesspringer4709
    @charlesspringer4709 6 місяців тому +21

    Nice. Presentation tip: Never use the word "what" unless you are asking a question. Eliminate "what we are going to do is we are going to" and "what we are going to do" and "we are going to". Also good to eliminate their friends "What you wanna do now is you wanna", etc. It takes some practice. I did it while teaching, so anybody can. "The last thing we are going to do now is we are going to..." ==> "Now..."

    • @Ordnas95
      @Ordnas95 6 місяців тому +2

      Thank you Charles

    • @zillibran
      @zillibran 6 місяців тому

      he said "waa" or "wa" has in y, IDIOT

    • @zillibran
      @zillibran 6 місяців тому

      ​​@@Ordnas95 he said "waa" or "wai" has in y

    • @zillibran
      @zillibran 6 місяців тому

      he said "waa" or "wa" has in y

  • @cribless810
    @cribless810 6 місяців тому +101

    Man youre getting spammed with "please solve jee questions now" so sad

    • @tmrapper6378
      @tmrapper6378 6 місяців тому

      He'll benefit from that

    • @ShanBojack
      @ShanBojack 3 місяці тому +2

      Jee is overrated fuck it

    • @tmrapper6378
      @tmrapper6378 3 місяці тому

      @@ShanBojack elaborate please

    • @Redamyx11
      @Redamyx11 3 місяці тому

      It indeed is overrated. We indians have to stop stamping our demands on youtubers to solve jee questions. We should suggest but PPL often keep crying , that makes me sad ​@@tmrapper6378

  • @elibrahimi1169
    @elibrahimi1169 6 місяців тому +17

    6:13 couldn't we just integrate both sides from the very beginning and get the same result ?

    • @jameslawson9754
      @jameslawson9754 6 місяців тому

      we dont know what the integral of y’ or y’’ with respect to y is

    • @elibrahimi1169
      @elibrahimi1169 6 місяців тому +3

      @@jameslawson9754 the i integral of yy' is 1/2 * y² and that of y" is y', and add a constant

    • @lucasredondo4234
      @lucasredondo4234 6 місяців тому +1

      @@elibrahimi1169the integral of y dy is y^2/2 , not the integral of y dy/dx , dy/dx being y’

    • @elibrahimi1169
      @elibrahimi1169 6 місяців тому +3

      @@lucasredondo4234 if you differentiate 1/2* y² you get yy' by the chain rule and the power rule, by that logic the antiderivative of yy' is 1/2 y² + C

    • @elibrahimi1169
      @elibrahimi1169 6 місяців тому +1

      @@lucasredondo4234 alright let's go with your logic ;
      the integral has a built in dx in it so the integral of int y dy/dx dx=int y dy =1/2 y² + C

  • @stevenwilson5556
    @stevenwilson5556 6 місяців тому

    10:36 "this is the beauty of differentials and integration, you never know what the fuck you're going to get"

  • @hornkneeeee
    @hornkneeeee 6 місяців тому +6

    i saw that y=0 holds for this equation then realised any constant would work, Proof by observation as the experts call it

  • @TrailersReheard
    @TrailersReheard 6 місяців тому +7

    THIS IS WHAT IM TALKIN ABOUT WAHOO

  • @constantinzimmer1747
    @constantinzimmer1747 6 місяців тому +13

    also y =-2/(x+k) for c = 0 right?

    • @dank.
      @dank. 6 місяців тому

      For c=0 you get the y=0 case, but for values of c very close to 0, yes.

    • @DangiMiner
      @DangiMiner 6 місяців тому +3

      ​@@dank.Actually he's right. In case c=0 you get this solution, not y=0.

  • @wyattstevens8574
    @wyattstevens8574 Місяць тому

    The form (except when dy/dx=0) is just "what function(s) is/are the logarithmic derivatives of their *own* derivatives?"

  • @garnetg987
    @garnetg987 6 місяців тому

    That opening meme about the 347% error made me burst out in laughter.
    Great video.

  • @renesperb
    @renesperb 6 місяців тому

    If you see that (1/2y[x]^2)' = y[x]*y'[x] you find that y ' =1/2 y^2+const. ,which is a separable differential equation , and it is easy to solve.

  • @rastakov1899
    @rastakov1899 6 місяців тому

    Thank you for your video.

  • @loliamtheone5643
    @loliamtheone5643 5 місяців тому +2

    just 1 quick question why not integration by parts ? is it wrong i have no clue here...with respect to y if we integrate the entire function ie both sides we get a simpler version of the equation which he initially established
    no one is seeing this but im 1000% making this differential integtration un nuts

  • @falquicao8331
    @falquicao8331 6 місяців тому

    There are two missing solutions for c=0,
    y = -2/(x+r)
    And c

  • @XanderGouws
    @XanderGouws 6 місяців тому

    My favorite solution is -2/x >:) It's easy enough to show its a solution by plugging it into the initial equation.
    Though it's not explicitly in the class of solutions you found, you get it when you take c->0 asymptotically and constrain kappa = sqrt(2c) * pi/2.

  • @senco445
    @senco445 6 місяців тому

    Fabulous! This was highly entertaining!

  • @ImLucky6
    @ImLucky6 6 місяців тому

    i would probably start by converting these into mcclaurn series, and then just pray from there

  • @speadskater
    @speadskater 6 місяців тому +1

    This could be written as an exponential

  • @tszhanglau5747
    @tszhanglau5747 6 місяців тому +1

    should have titled this "Deez Differential Equation is Nuts"

  • @brickie9816
    @brickie9816 6 місяців тому

    i approached it like so:
    yy' = y''
    2yy' = 2y''
    yy' + yy' = 2y''
    (y*y)' = 2y'' (power rule)
    y^2 = 2y' (integrate both sides)
    y^2/2 = y'
    i have arrived at the same expression as you but didn't know how to continue so i very much appreciate the video!

    • @ntuneric
      @ntuneric 6 місяців тому

      +C

    • @brickie9816
      @brickie9816 6 місяців тому

      @@ntuneric good catch, idk how I could have forgotten

  • @akirakato1293
    @akirakato1293 3 місяці тому

    with your substitution of t(y)=y'(x), when can you make such a substitution (you assume y'(x) can be a function of y alone)? is it because the differential equation only involves differentials of y?

  • @PixelSergey
    @PixelSergey 6 місяців тому

    You said that (y-dt/dy)*t=0 for all x implies that (y-dt/dy)=0 for all x or t=0 for all x. But surely one can be zero on some interval and the other can be zero on the other intervals? 4:02

  • @TNT9182-j1e
    @TNT9182-j1e 6 місяців тому +2

    I actually got one of these right!?!?!?!?!? wow.
    I just noticed that d/dx(y^2) = 2y dy/dx

  • @tusharsr2709
    @tusharsr2709 6 місяців тому

    the base question was integrable in itself, you'd get the same result

  • @MauriceL2006
    @MauriceL2006 6 місяців тому

    Differential equations are fun that by some nontrivial transformation sometimes we can reduce the equation we know how to solve ❤

  • @vatsalyavashisth828
    @vatsalyavashisth828 6 місяців тому +1

    great video !!!
    10:39 XD

  • @adityamishra0706
    @adityamishra0706 6 місяців тому +2

    It's good

  • @هالهالبكري-م6ص
    @هالهالبكري-م6ص 6 місяців тому +1

    yy'=y"
    1/2Dy^2=Dy'
    1/2y^2=y'
    Int dx=Int 2/y^2 dy
    x +c =2/y
    y=2/x+c

  • @samueldeandrade8535
    @samueldeandrade8535 6 місяців тому +1

    It seems someone wrote
    yy' = y''
    instead of
    $yy' = y''$

  • @Anonymous-zp4hb
    @Anonymous-zp4hb 5 місяців тому

    Uh oh. I got a different answer but can't see the flaw in my reasoning:
    yy' = y"
    Integration by parts:
    uv = $(uv')dx + $(vu')dx
    let u = v = y
    then:
    yy = 2 $(yy')dx = 2y'
    (d/dx)(1/y(x))
    = -y'/yy [chain rule]
    = -1/2 [substitute above identity]
    If (d/dx)(1/y) is a constant, then
    y = 1/(mx+c)
    And..
    y' = -m/(mx+c)(mx+c)
    y" = 2mm/(mx+c)(mx+c)(mx+c)
    if yy' = y"
    then 2mm = -m
    i.e. m = 0 OR m = -1/2
    So the only 2 function families that solve this differential equation
    should be:
    1.) y = c where c can be any real number.
    2.) y = -2/(x+c) where c can be any real number.
    Edit: I've spotted the mistake.
    I missed the constant of integration. Including it means (d/dx)(1/y) is not a constant like I said it is.

  • @JacquesRGAO
    @JacquesRGAO 6 місяців тому

    yy'=1/2(f^2)', thus $\frac{f^2}{2}-f'=C$, now this is a separable equation, something like $f=k\frac{Ae^{kx}+1}{Ae^{kx}-1}$...

  • @ibnSafaa
    @ibnSafaa 6 місяців тому

    You can rewrite it as :
    y = C.tan(Ax+B)

    • @julianbruns7459
      @julianbruns7459 4 місяці тому

      The final solution only has 2 free parameters while yours has 3. (C and A are not independent from each other) it would rather be 2A*tan(Ax+B)

  • @denizgoksu9868
    @denizgoksu9868 6 місяців тому

    I like that schizo chain rule way of transforming the equation

  • @mastershooter64
    @mastershooter64 6 місяців тому +2

    plz eolve JBEE Retreat questions they arecthe heardest examn in de world1!1!!!1!1!1!!!1!1!!!
    Jokes aside that is a pretty crazy ODE

  • @Oricorio-Pom-Pom
    @Oricorio-Pom-Pom 6 місяців тому

    y•y^’=y^(‘•’)
    y^(‘+1)=y^(‘^2)
    log_y both sides
    ‘+1=‘^2
    -(‘),-1 both sides
    0=(‘^2)-(‘)-1
    >is quadratic in form a=1, b=(-1), c=(-1)
    => (1(+/-)sqrt(1-(-4))/2
    => (1(+/-)sqrt(5))/2
    Positive happens to work out to (‘)= golden ratio
    Negative is (1-sqrt(5))/2
    >cursed, but cool coincidence.
    Final answer; (‘) = φ, (1-sqrt(5))/2

  • @brandonklein1
    @brandonklein1 6 місяців тому

    You missed an interesting solution if C=0 though! If c=0 y=-1/x, which is also a solution.

  • @Wielorybkek
    @Wielorybkek 6 місяців тому

    neat!

  • @KojackD1B
    @KojackD1B 6 місяців тому

    You forgot the solution when c=0 not just when c=! 0 ❤

  • @samueldeandrade8535
    @samueldeandrade8535 6 місяців тому +1

    Flammy looking hot, in more than one way. We gonna have shirtless video one day? ❤

  • @machoodin5172
    @machoodin5172 6 місяців тому +1

    Why c cant be 0? If c was zero the integral would have another solution

  • @Himanish-wu6ip
    @Himanish-wu6ip 6 місяців тому

    I think another solution is y=-2/x

  • @kappasphere
    @kappasphere 6 місяців тому

    My approach:
    y y' = y''
    (y²)'/2 = y''
    Integrate both sides (I'm ignoring the +C because the problem will be annoying, I'm not doing partial fractal decomposition) :
    y³/6 = dy/dx
    Separation of variables:
    int dx = 6 int 1/y³ dy
    x = 6/(-2) * 1/y² + A
    x = A - 3/y²
    1/y² = A - x/3
    y = 1/sqrt(A - x/3)

  • @sans1331
    @sans1331 6 місяців тому

    simple.
    yy’=y’’
    raise both sides to the power of 1/‘
    yy=y’
    y^2=y’
    ‘=2
    edit: this is a joke btw lmao

    • @Oricorio-Pom-Pom
      @Oricorio-Pom-Pom 6 місяців тому

      If we still take “‘“ as a variable, the actual answer is that ‘ has two solutions, one of which is the golden ratio, which i find even more cursed than what this is.

    • @Oricorio-Pom-Pom
      @Oricorio-Pom-Pom 6 місяців тому

      If ‘=2, y^(2+1)=y^(2^2)
      Simplifies to y^3=y^4, which is only true at y=0

  • @nif4345
    @nif4345 6 місяців тому

    y=0

  • @victory6468
    @victory6468 6 місяців тому

    I got -2/x

  • @johnwt7333
    @johnwt7333 6 місяців тому +1

    Shirtless video soon?

  • @mizjackyhoward3495
    @mizjackyhoward3495 6 місяців тому

    Woohoo

  • @ofridaniel2127
    @ofridaniel2127 6 місяців тому

    2 views in 3 seconds bro has fell off

  • @adityamishra0706
    @adityamishra0706 6 місяців тому +3

    Plz solve jee advanced questions ❤

  • @tophatjones6241
    @tophatjones6241 6 місяців тому

    Please Solve JEE Advanced Questions

    • @virat.chauhan
      @virat.chauhan 6 місяців тому +12

      Motivate him to solve something serious i.e. Millennium Problems

    • @jorex6816
      @jorex6816 6 місяців тому +3

      ​@@virat.chauhan haha

  • @ntuneric
    @ntuneric 6 місяців тому

    i did it as y'=p(y)
    y''=(p(y))'=p'y'=p'p. here p'=dp/dy
    yp=p'p
    case 1. p=0 → y'=0 → y=constant
    case 2. p≠0 → p'=y → p=½y²+k
    from here y'/(½y²+k)=1
    if k > 0, k=2A² → y = 2A·tan(Ax+B)
    if k = 0 → y=-2/(x+C)
    if k < 0, k=-2A² → -2A·tanh(Ax+B)

    • @ntuneric
      @ntuneric 6 місяців тому

      (the video told me to post my solution before watching so i didnt know i would have the same approach
      also peeprimepee peepeeprime)