Newton's method (introduction & example)
Рет қаралды 182,970
Күн бұрынLearn more than just Newton's method on Brilliant 👉 brilliant.org/blackpenredpen/ (20% off with this link!)
Using Newton’s method to solve a quintic equation! Newton's method is one of the must-know topics in calculus 1 and the concept is just based on using the equations of the tangent lines.
Check out my playlist on @bprpcalculusbasics for more Newton's method worked-out examples: 👉 • Newton's Method, calcu...
💪 Join our channel membership to unlock special perks,: bit.ly/34PGH4h
🛍 Shop math t-shirts & hoodies: bit.ly/bprpmerch
10% off with the code "WELCOME10"
----------------------------------------
**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 Jude Khine R3factor. Jasmine
----------------------------------------
💪 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
#bprp
0:00 opening story
2:05 deriving Newton's method
10:07 using Newton's method to "solve" the quintic equation
19:49 check out Brilliant to learn more calculus!
20:47 Fun fact, x^5-5x+3 is actually factorable
Try "my first quintic equation" here 👉 • My First Quintic Equation
@blackpenredpen 2 жыл бұрын
Fun fact, x^5-5x+3 is actually factorable and we will factor it soon! Try "my first quintic equation" here 👉 kzfaq.info/get/bejne/j9F1g7CXltKWn6c.html
@MathNerd1729 2 жыл бұрын
You're very sneaky with the surprise! 1/phi and -phi are the two other roots! I found them by trying to factor x⁵ - 5x + 3 via trial and error and finding (x² + x - 1)(x³ - x² + 2x - 3) = 0 This somewhat reminds me of generating functions, specifically this result involving the Fibonacci numbers: 1/(1 - x - x²) = 1 + x + 2x² + 3x³ + 5x⁴ + 8x⁵ + . . . By the way, I used the cubic formula on the second factor; the root you found in the video can be written as such: ⅙ × (2 + ∛(260 + 4 √4725) + ∛(260 - 4 √4725)) Math is fun! :)
@holyshit922 2 жыл бұрын
Yeah , number opposite to golden ratio and reciprocal of golden ratio , there is also third one but i didnt recognize it Two other roots are complex x^5-5x+3=(x^3 - x^2 + 2x - 3)(x^2 + x - 1)
@blackpenredpen 2 жыл бұрын
@@MathNerd1729 👍 And I am glad that you enjoyed it too. This is just a warm up to a video coming in a week or two. Btw I used the Lagrange resolvent for that cubic.
@MathNerd1729 2 жыл бұрын
@@blackpenredpen Nice! I just massaged it into a depressed cubic then used "Cardano's" Formula since I couldn't be bothered to remember Lagrange's resolvent. I made a slight mistake in depressing the cubic on my first attempt though, so maybe it would've been better to use Lagrange. 😂
@aashsyed1277 2 жыл бұрын
Every quintic is factorable since it has zeroes!!
@tonmoisingh8197 2 жыл бұрын
man i absolutely hate it when someone jumps out of the blue and asks me to solve a quintic equation :/
@chitlitlah 2 жыл бұрын
I had that happen at the supermarket last week. My milk spoiled and my eggs hatched while I was helping him work it out.
@landsgevaer 2 жыл бұрын
Were it a lady, I would be very much attracted to her...
@Lolwutdesu9000 2 жыл бұрын
Indeed, tfw.
@rayquaza5908 2 жыл бұрын
I was unable to solve it and now I am chilling with Satan (●__●).
@fantiscious 2 жыл бұрын
I've learned how to solve cubics recently. NEVER trying quartics...
@fantiscious 2 жыл бұрын
15:19 "Everything is doable if you have patience. But I have a calculator" *Proof mathematicians are not accountants*
@ronaldrosete4086 2 жыл бұрын
"Everything is doable if we have patience." -BlackPenRedPen, 2022 15:19
@blackpenredpen 2 жыл бұрын
lol : )
@freemanbryan8873 Жыл бұрын
Just realized sometimes maths is hard because we aren't taught the basics like this. My Lecturer did this whole chapter without drawing a single graph.
@InXLsisDeo Жыл бұрын
Pretty bad teacher, tbh. Every book that exposes the Newton's method has a graph that makes it limpid how it works. I'm pretty sure Newton himself has a graph of it in his publication. Wikipedia has a nice animation
@saujanyapoudel8910 7 ай бұрын
And some teachers just 'Memorise' everything and expect you to follow along word by word. My teacher told me its wrong to use notations of xn-1 and xn instead of xn and xn+1 cause according to him n starts at 0 and my notation would give a negative symbol for x. He walked away even before I could argue how starting n at 1 would fix everything 😅
@Kk_brayann 7 ай бұрын
Intact 😢
@MikehMike01 4 ай бұрын
Teachers are mostly incompetent
@barobabu8889 2 жыл бұрын
The other two solutions are x = -1.618034 and x = 0.618034 And I see why you said the answers would be interesting !!
@TechTactics4253 2 жыл бұрын
golden ratio and its inverse?
@_miobrot_603 Жыл бұрын
negative golden ratio and positive reciprocal
@tomi6701 Жыл бұрын
the fact that he manages to explain everything and write it all on one small board is amazing!
@diogoprudente6041 2 жыл бұрын
im almost graduating at mathematics and i cant express how much i love your yt channel
@bradryan8071 2 жыл бұрын
Just taught this to my grade 11's this week. They loved how we can finally solve equations involving different classes of functions, where no closed form solutions are possible. The best part of Newton's Method is that with a good "First Guess" the solution converges very quickly to remarkable accuracy. We also discussed and I showed them graphically, what happens when you do not start off with a good initial value.
@NoNameAtAll2 2 жыл бұрын
have you shown them 3blue1brown's video on Newton's fractal?
@bradryan8071 2 жыл бұрын
@@NoNameAtAll2 In getting ready for AP exams and working through the curriculum, I find that I have very little time to do anything but teach concepts and then solve as many different types of problems as possible. This is not a complaint, rather, the joy of what we are discussing is in the pure application of using these great tricks to actually work out real world useful problems.
@arequina 2 жыл бұрын
the "first guess" is the crucial part.
@ramansb8924 2 жыл бұрын
@@NoNameAtAll2 i have, it was mind blowing 🤯🤯🤯
@okaro6595 2 жыл бұрын
Newton's method is easy to do with a calculator. First enter the initial guess and press = (or Exe on some models). Then enter the x-f(x)/f'(x) using Ans in place of x. Then you just hit = until the value does not change. Many calculators have built in Newton's method. I tried to trick it by giving initial value where f'(x)=0 but it did not fool. They likely use some small difference in x instead of a true derivate. The nice thing in Newton's method is that even if you make an error it likely just slows you down, you will not get a wrong answer.
@areejmakhamra4016 7 ай бұрын
The explanation is very beautiful. Thank you for these efforts🤍🤍
@alberteinstein3612 2 жыл бұрын
Newton, what a legend 😎
@blackpenredpen 2 жыл бұрын
So are you, Mr Einstein!
@alberteinstein3612 2 жыл бұрын
@@blackpenredpen thanks! :)
@user-td9nx4xo6u Ай бұрын
Thanks for this video, it helps me to understand the deduction part of this Newton-Raphson method, and it’s linked to the straight line equation unexpectedly. Wow!
@troyshrauger3576 2 жыл бұрын
Literally just co-wrote a paper on Galois. I love this video so much.
@ampisiades 2 жыл бұрын
Congrats for the 903k followers. You will get a million just fast!!!
@sm-pz8er 6 ай бұрын
Very well explained. Thank you so much
@ehoumanjohann543 2 жыл бұрын
Thanks for this video , I have a better un comprehension of this optimization’s method
@ffggddss 2 жыл бұрын
f(x) = x⁵ - 5x + 3 = 0 f'(x) = 5(x⁴ - 1) I get x = 1.275682204... = R₁ for that first solution you got (the largest of the 3 real roots); we agree on that. For the next one to the left, I get x = .6180339887... = R₂ And for the smallest of the 3 real roots, I get x = -1.618033989... = R₃ , and I recognize these as R₂ = 1/ϕ [= ϕ-1] and R₃ = -ϕ. This means that g(x) = x² + x - 1 is a factor of f(x). [The monic quadratic in x, with sum-of-zeros = S, and product-of-zeros = P, is x² - Sx + P.] Dividing, we get f(x)/g(x) = x³ - x² + 2x - 3 = h(x) whose only real zero must be R₁ . My calculator confirms this. At this point, you could get down to a quadratic (which will have to have a discriminant < 0), by dividing: h(x)/(x-R₁) = x² + bx + c = q(x) which we can find by noting that multiplying (x-R₁)q(x) and setting it equal to h(x), tells us that b - R₁ = -1; and R₁c = 3. Thus, b = R₁ - 1 = .275682204... c = -3/R₁ = -2.351682881... So finally, (R₄ , R₅) = -.137841102... ± (1.527312251...)i Fred PS. Zeros R₂ and R₃ can be verified using the relation of Fibonacci numbers to powers of ϕ: ϕⁿ = F[n]ϕ + F[n-1] R₂ = 1/ϕ = ϕ⁻⁻¹ = ϕ-1 R₂⁵ = ϕ⁻⁻⁵ = F[-5]ϕ + F[-6] = 5ϕ - 8 f(R₂) = R₂⁵ - 5R₂ + 3 = 5ϕ - 8 - 5ϕ + 5 + 3 = 0 R₃ = -ϕ R₃⁵ = -ϕ⁵ = -F[5]ϕ - F[4] = -5ϕ - 3 f(R₃) = R₃⁵ - 5R₃ + 3 = -5ϕ - 3 - 5(-ϕ) + 3 = 0
@muhammadarfighifari8292 2 жыл бұрын
Holly schiedd....
@Pedro-yv1kk 2 жыл бұрын
Congrats, you are too smart. But would be cool to get the exact algebric solutions with the cubic equation that you finded, this w'ont be easy, if I get time I'll do this.
@ffggddss 2 жыл бұрын
@@Pedro-yv1kk Spoiler Alert: It's already been done by Math Nerd 1729, in a reply to bprp's self-pinned comment. His/her real zero simplifies to: ⅙[2 + ∛(60√21 + 260) - ∛(60√21 - 260)] That cubic being h(x) = x³ - x² + 2x - 3 Math Nerd 1729 doesn't provide the complex conjugate pair of zeros, but if you're better versed in the cubic solution formula than I am, maybe you can deduce them from the information above. Fred
@maximilianthegreatest7763 2 жыл бұрын
The average of the solutions is 0. Why does this happen?
@ffggddss 2 жыл бұрын
@@maximilianthegreatest7763 The sum of all n zeros of an n-degree polynomial is always -b/a, where a is the leading coefficient and b is the next one; the (n-1)st power. So whenever the (n-1)-power term vanishes, the sum of solutions is 0. And of course, so is the average, which = sum/n. Fred
@yashrajnanda3199 2 жыл бұрын
Enjoyed to see your video with help me in better understanding of maths Through your videos it make me feel that maths in flowing in my vein Thanks to give thie type of content
@faisal_tarshon 10 ай бұрын
Thank u man , u make math fun and easy
@DrLiangMath 2 жыл бұрын
Excellent lesson for Newton's method!
@davidgould9431 2 жыл бұрын
Circa 5:00 Rather than calculating the equation of the tangent line, you only need the slope, ie the "rise over run", so f'(x₁) = f(x₁)/(x₁ - x₂) which trivially rearranges to x₂ = x₁ - f(x₁)/f'(x₁) It's basically the same thing, but I think it's easier to explain. Edit: forgot my ₁s inside the f() f'()
@goldfing5898 Жыл бұрын
Very good explanations, but two things: 1.) I would improve the sketch by adding vertical lines, e.g. from (x1, 0) to (x1, f(x1)), not only the tangents, so you can see a zig-zag pattern in the sketch, which better demonstrates the iteration process. 2.) I learned the formula x_{n+1} = x_n - f(x_n)/f'(x_n) which is easier to write down and memorize than x_n = x_{n-1} - f(x_{n-1}) / f'(x_{n-1}) due to the shorter indices.
@TheGunser98 Жыл бұрын
I used this method for a subject at my university, It's wonderful! can you do the same thing with Lagrange?
@letstalksciencewithshashwa9527 6 ай бұрын
He's the guy Steven's dad compares his son with
@lanjiaxu9531 2 ай бұрын
Thanks a lot!
@user-rm8yz2hw9i 6 ай бұрын
Great job ❤🎉
@fiona685 Жыл бұрын
Thank you!
@brumarul7481 2 жыл бұрын
I made an algorithm now that uses this method, ty for showing it.
@trentrodgers1875 9 ай бұрын
Thank you legend !
@ipcheng8022 2 жыл бұрын
I just did a tutorial about newton method for UG students last week
@blackpenredpen 2 жыл бұрын
Perfect!
@leif1075 2 жыл бұрын
@@blackpenredpen why would anyone think to use the intermediate value theorem? I don't see why anyone would? Hope you can respond when you can.
@betatesting2052 2 жыл бұрын
x^5 - 5x + 3 can also be solved using a recursive definition for x. x^5 = 5x - 3 x = (5x-3)^(1/5) Now, replace the value of x with (5x-3)^(1/5) x = 5(5x-3)^(1/5)+3)^(1/5) The idea is that if you substitute this enough times, eventually you can plug in ANY value for the x on the right hand side and it will still return one root of x, at which point you can use polynomial division to reduce it to a solveable quartic equation, as the more times the definition is plugged in, the more this calculation will tend to converge on a single value, the solution to x^5 - 5x + 3. In a way, this is like a recursive sequence, where u[n+1] = (5(u[n])+3)^(1/5). In fact, you can use this recursive method for any equation with 2 instances of the variable - I don't know if it could possibly be adapted for more than 2 values. Hope someone found this useful!
@anshumanagrawal346 2 жыл бұрын
You made a small algebraic error, that should be x^5 = 5x -3 , not plus 3
@betatesting2052 2 жыл бұрын
@@anshumanagrawal346 Thank you, I've corrected it now.
@goldfing5898 Жыл бұрын
You can also x^5 - 5x + 3 = 0 as 5x = x^5 + 3 x = (x^5 + 3)/5 and ist this as another Banach fixpoint iteration. For the initial value x0 = 0 or -1 or +1, it converges to 0.618... For x0 = 2 or -2, it diverges to infinite. If I remember correctly from the numerics lecture, if you write x = phi(x) then in order to converge, | phi'(x) |
@davidgillies620 2 жыл бұрын
Let p_n(x) be a trinomial of the form x^n + b x + c. If b is (-1)^n F_n and c is (-1)^(n - 1) F_n-1 with F the Fibonacci numbers, then p has a factor of x^2 + x - 1. So for example p_10(x) = x^10 + 55 x - 34 has factor x^2 + x - 1. The other factor is a polynomial with its coefficients taken from the Fibonacci series, in this case x^8 - x^7 + 2 x^6 - 3 x^5 + 5 x^4 - 8 x^3 + 13 x^2 - 21 x + 34. This ties in to the generating function of 1/(x^2 + x + 1).
@rusparmesan 2 жыл бұрын
I was surprised it took 7 iteration to get that approximation, I thought it would converge much faster, and 3-4th iteration will be already quite precise
@nononnomonohjghdgdshrsrhsjgd 3 ай бұрын
Perfect!
@templateaccount7816 2 жыл бұрын
We had some weird calculus aasignment in school where you were supposed to prove that the integral of -ax^2+b from root to root always was 3/2. I thought it didnt make any sense
@flowingafterglow629 2 жыл бұрын
OK, is there any way to set this up to solve x as the limit of when xn - xn-1 goes to 0? I guess for the other solutions start at 0 and -2. That should get there, right? There is a solution between 0 (pos) and 1 (neg) and a solution between -1 (pos) and -2 (neg), but -1 will blow up so start at -2
@diceman7155 2 жыл бұрын
Give us video on finite series please
@IbrahimNakshbndi 2 жыл бұрын
nice work 👑 am introducing mathematics in simple ways
@davidjames1684 2 жыл бұрын
This is very easy to approximate just by inspection. Use 0.6 since the x^5 term will be very small, and -5x will be -3, thus "almost" solving the equation. From there, I can just check values close to 0.6 on a computer, which I did, and quickly got about 0.618034.
@YoutubeBS Жыл бұрын
It's phi-1
@kfibcudwiefjw7428 Жыл бұрын
Amazing!
@Jkauppa Жыл бұрын
try consecutive derivatives equation = 0 (zero points of the equation nth-derivatives) to get the newton's method initial guess x0 regions directly
@user-oz5hi1px7e 2 жыл бұрын
i was talking to my friend about this today, and you released a video on it. coincidence 😄
@blackpenredpen 2 жыл бұрын
😆
@chitlitlah 2 жыл бұрын
It's not a coincidence. BPRP is watching you.
@Infinium 2 жыл бұрын
I’m in the middle of making a video on the Newton-Raphson method! ❤️
@nzbdv 2 ай бұрын
I just calculated solutions of quadratic using this method and rounding up to sixth decimal place and compared it with delta results and they were extremely close to each other. Newton was a genius by inventing this, and he was doing all his calculations without computers which is even more insane
@joaolima4787 2 жыл бұрын
Like we do in derivatives, when we take the limit as t goes to infinite to find the derivative, couldn't we just d the same and put in some calculator this iteration and make it goes to infinity to find the exact answer?
@tentanghukumkita6381 2 жыл бұрын
use newton's method, no quintic formula Detailed and interesting explanation Thank you 👍
@alextskm 2 жыл бұрын
Is there a way where we can do this n times and take n->inf , so we can get the exact solution or at least almost the exact solution ?
@watching4410 Жыл бұрын
This is best video about newton's method I didn't understand how got formula y-f(x_1) = f`(x_1) (x-x_1) Some else mentioned right angle to drive the equation, maybe that video explained how derive equation. He really used decimal value 1.67 instead of 1 2/3 or 5/3
@user-td9nx4xo6u Ай бұрын
This formula is according to the straight line equation，y-y1=k(x-x1). Just putting the f(x1) as y1, the derivative of the function as k, and x1 as x1, which the blogger has mentioned in the video. And I suppose you could understand better now. 😊
@namumakwembo Жыл бұрын
wow , thanks , now i know where the formula comes from
@createyourownfuture5410 2 жыл бұрын
16:48" I don't have enough space, so just trust me on that." Now where have I heard that before?
@darshandesai7225 2 жыл бұрын
The only way for newtons method to work is if f(x) does not change slope on the interval from the actual x value of the solution to the initial guess x1 right?
@leolesnjakovic8725 2 жыл бұрын
VERY INFORMATION VIDEO!!! TANK YU!! (soryr for bad egluish)
@ftbex9224 2 жыл бұрын
老師請問您之前說要拍有關beta function的主題，可是我找不到耶。
@DefenderTerrarian 2 жыл бұрын
I just got done with SecMath3. How does one calculate the derivative?
@onlythefacts999 2 жыл бұрын
Newton's method is also one of the easiest (or at least most well known) methods for solving Kepler's equation for the eccentric anomaly.
@purnaprajnabangere9820 Жыл бұрын
Can you please do a video on Galois Theory?
@i2italianialcollege 9 ай бұрын
te amo cineseeeeeeeeeeeeeeeee, so much love for youuuuuuu
@jimScienceNerd Жыл бұрын
I have used this at work to find roots and called it Newton iteration
@nkululekongidi6085 Жыл бұрын
will I be correct if I use my X1=1,5??
@sr.tarsaimsingh9294 2 жыл бұрын
00:46 we don't have to worry ; Because we will learn how to manage with such problem in next 20min.😁
@tobybartels8426 2 жыл бұрын
But we *do* have formulas for 5th and higher degree polynomials! It's just that the formulas involve operations slightly more complicated than roots. You say that you can solve a quadratic polynomial by completing the square or using the quadratic formula, but to do this, you have to introduce this crazy square-root operation. How do you define that?, and how do you calculate that? You define a square root as a solution to a particular equation, pretty circular if you ask me. And you can calculate it by hand using a technique that goes back to the early days of Algebra but is basically Newton's Method in disguise. And so, when the time comes to solve arbitrary quintic polynomials, you can define a new operation, slightly more complicated than the 5th root, as the solution to a particular equation. With the help of that, there's a formula to solve any 5th-degree polynomial (even ones that can't be factored). And you can go on and do the 6th degree and higher with the help of additional operations. Just like with the square root, these operations are all defined as the solutions to certain polynomial equations, and they can be calculated by hand to arbitrary precision using techniques that are basically Newton's Method.
@georget8008 2 жыл бұрын
Actually, when I was taught this method, it was named the Newton-Raphson method.
@AchtungBaby77 2 жыл бұрын
The two names are interchangeable. Plain old Newton's Method uses less syllables 😅
@ffggddss 2 жыл бұрын
When I was taught it, it was called just, "Newton's Method." Later I started seeing it called the "Newton-Raphson Method." I guess I'm in Achtung Baby's camp; when I refer to it, I just go with the shorter name. Fred
@sufma6505 9 ай бұрын
the legend
@kunwarpartapbhattal5015 2 жыл бұрын
What is coefficient of x^3n in expension of (1+x+x²+x³......x^2n)³
@slarx2476 8 ай бұрын
I noticed that if you choose a guess close to that local max on the side of a board on the left 3:14, then the tangent line’s x-intercept can overgo the answer, i think if it does that we need to continue to work with that guees, Id like to see a formula for a relation of the slope of the guess point and the distance of a guess from the solution TO the interval when tangent line overgoing the answer, and if this formula exists than the first point at which tangent line over goes the answer would be the answer. In other words I described the point whos tangent line solve newton’s method in just one step. What happens when x-intercept of a tangent line over goes the answer? Do we continue by starting again with that acquired guess, i can think that we now go in the opposite direction. i think because formula works with different slopes, its should be ready for the to slope going in a different direction. That is equivalent to starting Newton’s method all over again with that acquired overflown guess point
@holyshit922 2 жыл бұрын
Speaking the numerical methods did you try calculating polynomial roots using eigenvalue methods Two of the companion matrices are already in Hessenberg form Suggested method for calculating eigenvalues is QR method with shifts To QR decomposition we can multiply matrix A by rotation matrices xor by reflection matrices I didnt find how to choose clever shift Double implicit shift is tempting because entries of matrix are real but eigenvalues may be complex
@onlythefacts999 2 жыл бұрын
Still haven't found a great video on the doubly implicit shifted QR algorithm, I'd love to see one some day. Pretty rough way to calculate polynomial roots, though you get all the values in one algorithm at least (and eigenvalues/eigenvectors for nearly any complex square matrix on top of that). Edit: And no need for derivatives either!
@holyshit922 2 жыл бұрын
If you want to approximate roots of polynomial x^n=c_{1}x^{n-1}+c_{2}x^{n-2}+...+c_{n} with eigenvalues method you place coefficients c_{1}, c_{2},...,c_{n} in the first row xor last column , put ones just below main diagonal and everywhere else put zero In this way you will create companion matrix To get QR decomposition we can use multiplication by rotation matrices xor multiplication by reflection matrices Basic QR algorithm looks as follows A_{i}=Q_{i}R_{i} A_{i+1}=R_{i}Q_{i} but for faster convergence clever choice of shift is helpful For me doubly implicit shift is tempting because matrix entries are real and eigenvalues may be complex but there are other strategies for choosing shift
@onlythefacts999 2 жыл бұрын
Yeah, regular QR algorithm is pretty simple, it's the implicit double shift that I struggle with. Trying to find the submatricies of the Shur form without explicitly forming the matrix is where my brain shuts down. I said it's a rough way to find the polynomial roots because you need the double shift to find complex roots using QR while Newton's method can find all complex roots with good enough starting points.
@holyshit922 2 жыл бұрын
@@onlythefacts999 I found some examples that unshifed QR method doesnt converge to Schur form (fe matrix similar to the companion matrix of odd or even polynomial) (real eigenvalues on diagonal , complex eigenvalues from 2x2 block on diagonal) Moreover there can be found matrices for which convergence is very slow
@holyshit922 2 жыл бұрын
@@onlythefacts999 Have you read Numerical recipes in C ? There is an example of qr method with Hessenberg reduction , shifts and deflation
@framedthunder6436 5 ай бұрын
1:20 there is a quintic formula using series
@KhakiShow 4 ай бұрын
thxxxx
@PatrickxJames 11 ай бұрын
Finding all solutions here is simple: just find x values where the derivative is zero and run Newtons method at most once on both sides of the x values found. For instance, in this example we have f’(x) = 5x^4 - 5. Setting f’(x)=0 gives x=1 and x=-1 as solutions where the slope is 0. This means that there will be 3 solutions, since there are 3 different intervals where the slope does not equal 0: [-inf, -1), (-1, 1), and (1, +inf). Now just pick points within each interval and run Newton’s method to find the 3 solutions.
@meaghanjakobi4826 2 жыл бұрын
Can you make a rhénan sum or limit to go faster?
@reidflemingworldstoughestm1394 2 жыл бұрын
So what happens when the root falls at an extremum of f(x)?
@narangfamily7668 5 күн бұрын
CRAZY!!!!!
@louis8551 14 сағат бұрын
Is this the same as the Newton raphson ietrativ Methode ?
@flowingafterglow629 2 жыл бұрын
How do we find the two complex roots? Find the reals and then polynomial division to get to a quadratic equation?
@danieltefera7347 2 жыл бұрын
Just put in a complex guess
@hilm6245 2 жыл бұрын
i just loked up the quartic formula and i dont even know how insane a person must be to memorize that
@pulvinarpulvinar6749 2 жыл бұрын
How do we get the basic formula at 5:00 ? I don't understand the thought process
@blackpenredpen 2 жыл бұрын
y-y1=m(x-x1) equation of the line with the given point (x1, y1) and slope m
@WagesOfDestruction 15 күн бұрын
you can use it to solve implicit equations often too
@unfall5521 2 жыл бұрын
Can you not take the limit of Xn as x goes to infinity? I think it is doable because you have Xn+1 = g(Xn) so you solve limit L=g(L) Where g(x)= x - f(x)/f'(x). You basically have to solve L= L -f(L)/f(L) Which is f(L)/f'(L) =0 And now that I wrote it I get why this is not helpful lol i just got back to f(x)=0
@siddhesh_sinha 2 жыл бұрын
finally, I see your video just after its release. btw my DP is made from a mathematical function (cause of your inspiration).
@gregchen4023 3 ай бұрын
I'm Chinese from Mainland China. I guess you are from Hongkong or Singapore. Wonderful lecture! Difficult problem made easy!
@matte14-70 2 жыл бұрын
Can you make a video using the quartic formula ?? 😬😬😬
@abemcg3803 2 жыл бұрын
I like that you hold a Pokémon ball plush the entire equation
@danieltefera7347 2 жыл бұрын
It's secretly his microphone 😯
@abemcg3803 2 жыл бұрын
@@danieltefera7347 or he’s probably teaching it to Pikachu and he can hear it from the Pokéball
@danieltefera7347 2 жыл бұрын
@@abemcg3803 lol
@koksem 2 жыл бұрын
Hey I was trying to calculate the area of sin wave under a line, I chose y = 0.5x for this one. I decided than to find interval at which y=sin(x) is below y= 0.5x herefore I tried solving something like sin(x) < 0.5x which is not solvable. My question is how to find the interval at which y=sin(x) is below y= 0.5x
@ffggddss 2 жыл бұрын
If you graph those, I think you'll find that the two functions intersect at the origin (of course), and somewhere between x = ½π and x = 2, and nowhere else x > 0. And because both are odd functions [ f(-x) = -f(x) ], there's another intersection < 0, matching the first one, but with the signs of x & y reversed. So the places where sin(x) ≤ ½x, are from that intersection where x < 0, up to (0,0), and from the +ve intersection to ∞. And that means you need to find where sin(x) = ½x < 0. That is, the negative zero of f(x) = sin(x) - ½x which can be computed using Newton's Method, f'(x) = cos(x) - ½ You could try as the initial guess, x[0] = -2; then iterate x[k+1] = x[k] - f(x[k])/f'(x[k]) Fred
@koksem 2 жыл бұрын
@@ffggddss hmm, interesting. Ty for answer.
@sxkjknjw2 Жыл бұрын
BPRP really pulled out a "trust me, bro" at 6:44 lol good vid nonetheless
@Nik-qh7cq 2 жыл бұрын
What if you take 3 derivatives at least and then see were the possible roots are and then use Newton in order to go close to the value
@TheRealHedgehogSonic Жыл бұрын
I like to call this method "Chain chugging." Because it's essentially plugging and chugging every solution into the same formula until it stops varying so much.
@professorpoke 2 жыл бұрын
"Take the derivative, Take the derivative, Take the derivative, Take the derivative, Take the derivative." ~BPRP, 2022
@kepler4192 2 жыл бұрын
I'm pretty sure you made a video about it either in another channel or even in this one
@blackpenredpen 2 жыл бұрын
Yes. An old one with the omega constant. But you will see why this thing again 😃
@kepler4192 2 жыл бұрын
@@blackpenredpen interesting
@ek_minute_ Жыл бұрын
I literally lost my ear drums when you said "suppose" In the beginning
@AmitDash 2 жыл бұрын
What about x^5 -3x +5 = 0 With x1 = 1
@sardarbekomurbekov1030 2 жыл бұрын
Newton is the real genius.
@foerza649 2 жыл бұрын
I want to see quarntic formula please
@oggybob9167 2 жыл бұрын
The Quartic formula is bigger than lord of the rings 💀
@alejandroperalta5664 2 жыл бұрын
Donde venden ese cuadro del numero neper???
@logannuculaj487 9 күн бұрын
What would you do if all the zeroes of a function are complex? Would you still use Newton’s method? Maybe using tangent parabolas instead of tangent lines to approximate?
@richiewitkowski7142 Жыл бұрын
I absolutely hate when im just walking on the side of the street and some random guy asks me how to do Newton's method But thankfully I dont have to worry about this anymore 😮💨
@fasebingterfe6354 2 жыл бұрын
Aw, man! I hate it when I'm walking down the street and some stranger approaches me with a quintic ecuation!
@nirajandata 2 жыл бұрын
waiting for your IMO related videos
@SteelM4chine 2 жыл бұрын
1:11 who did it to you?
Nikita Zdradovskiy
Рет қаралды 7 МЛН
3Blue1Brown
Рет қаралды 2,8 МЛН
Ellie Sleightholm
Рет қаралды 48 М.