QR Decomposition of a matrix and applications to least squares Check out my Orthogonality playlist: • Orthogonal sets Subscribe to my channel: / @drpeyam
You are on my elite teachers list: Sal from Khan academy, The organic chemistry tutor, Professor Dave, and now Dr Peyam. Hats off to all of you for making my university life easier. Massive respect and love for all of you
I've Always seen matrix decompositions (QR, LU) done with square matrix, it's a bit strange for me to see them in the nonsquare world. Also, I was told that Q must be Unitarian and Hermitian (I guess for IR orthogonal and symmetric would be fine) hence making QR only be possible in square matrices by definition. I wonder how much of a lie resides in what I just told
Gram Schmidt is good for paper and pencil calculations but i heard that it is numerically unstable and I should avoid it when I write program for QR decomposition Householder reflections or Givens rotations are better choice for those who want to write a program Silly , or maybe not in Householder reflections we need transpose of matrix to get Q and square matrices are easy to transpose in place Transpose of rectangular matrix also can be done in place but it is not so easy
Thanks this video series has really been helpful! Also, just wondering if you would be willing to share but I was wondering what watch you're wearing? I think it looks great
Hey, thanks for the amazing video. I just have one quick doubt. As you said Q is orthonormal matrix, then when I compute Q*Q' it does not give I. Please enlighten me, or am I misunderstanding something?
thanks for the video! but one question, even if R' is not invertible (meaning A has linear dependency) , there still solution for LS right? just lose one dimension. No?
What, if you define a kind of Matrix-Product with the compositon of the elements: A = { {a11(x), a12(x)},{a12(x), a22(x)} } and B = { {b11(x), b12(x)},{b12(x), b22(x)} } So that, A°B = { { a11(b11(x)) + a21(b12(x)), a12(b11(x)) + a22(b12(x)) }, { a11(b21(x)) + a21(b22(x)), a11(b21(x)) + a21(b22(x)) } } Linear Algebra is boring, because I never understood it well. Make more videos about this crazy fractional calculus stuff! Something like this: (d/dx)^f(x) x = f'(x), where f(x) is the order of the derivative. Or functions that transform other functios to their derivatives: g1(f(x)) = f'(x), g2(f(x))= f"(x), ... where gn depends of (d/dx)^n f. Then you could "maybe" generalize derivatives by these functions --> g0.3(f) = (d/dx)^0.3 f
Orthogonal Matrices are a bit strange anyway. I will never understand why you call a Matrix with orthoNORMAL column-vectors orthoGONAL and not orthonormal. Most of the time I love maths, but sometimes I hate it xD.
I agree with you. Luckily, there is a book where authors say orthonormal matrix. In my lectures I also say this. And there are other cases where I use the corresponding correct name. Similar, more terrible thing is 😄 "this infinite series is convergent". Math is nice, but the language created by persons is not necessarily correct.
This guy always leaves a smile on my face
3 hrs of lecture and i didnt understand a word.
5 mins of watching this video - and i undersand every thing !!
thank you.
You are on my elite teachers list: Sal from Khan academy, The organic chemistry tutor, Professor Dave, and now Dr Peyam. Hats off to all of you for making my university life easier. Massive respect and love for all of you
Add Dr Trefor Bazett in that list. :)
3blue1brown too
You sir are a genius. This linear algebra extravaganza was super helpful!
Glad you like it 😄
This is the funnest math tutorial video I've ever seen. You made the process seem so streamlined and easy, thank you!
Best QR decomposition video I've found. Well explained, straight to the point, easy to understand. Thank you.
That's awesome. I don't understand matrix and I don't understand english but with your funny explinations, I understand everything. Thank you so much.
the radiance of his positivity in his teachings make me love linear algebra XD
Thank you for explaining a simple topic as it should be explained, in a simple way. Great explanation!
man how did i not find you before, you're litteraly going to make me pass numerical analysis
This was expained so well that i understood despite i talk spanish and dont know even a little of english
Very well taught i came here after trying to follow other tutorials the pace was slow enough i could understand
Thank you!!!
thanks sir,you are best teacher sir
The best explanation given on this topic!
Im at the point of the semester where I need to take whatever this guy took
great explanation! loved your energy
Thank you sir, you just saved me countless hours of non effective study.
You’re welcome!!
Thank you so much, great video and a lot of positivity!
I have been struggling with this topic, you explained this so well. Thank you!
What a living legend... Amazing Peyam
omg you're so cute i can watch you teach all day
Thanks so much for explanation, It was clear and concise and to the point.
thank dr peyam! I really liked the extension to least squares in the second half of the video.
Wow best explanation and nice style of teaching. Very precise and easy to understand
I wanted SVD and two grid method too :)
Thanks alot!!!!!!!!!!!!!!!!!!!!! Prepare to hand in my homework set~~~~~~~
WOW THANK YOU DR. YOU ARE THE BEST
You're my hero!
I’m Captain Peyamerica! 🙂
God Bless you, man!
Thank you so much. This really helped my understanding of qt decomposition
This tutorial is amazing, thank you
I've Always seen matrix decompositions (QR, LU) done with square matrix, it's a bit strange for me to see them in the nonsquare world. Also, I was told that Q must be Unitarian and Hermitian (I guess for IR orthogonal and symmetric would be fine) hence making QR only be possible in square matrices by definition. I wonder how much of a lie resides in what I just told
Thanks a lot for your clear explanation!
And I thought it was something to do with QR codes :P
🤔😅😁😁😀
Thankyou for being successful in successfully wasting my time
Awww you’re welcome!!
And the next decomposition should be the very important SV decomposition which one typically uses in the matrix product state formalism. :D
Thank you so much. Very helpful
Glad it was helpful!
Your enthousiasm is amazing, the only thing that triggers me is that you put your line of your Q on the wrong side !
Enthusiasm increases by exp(2) when Watching it on 2x
omg, thank you a lot for your priceless knowlage
Great explanation, thanks!
u are like the bob ross of math :) thank you
Thanks a lot!! The example was very illustrating!
Perfect explanation. Thx professor
I love this man
absolutely fantastic
Gram Schmidt is good for paper and pencil calculations but i heard that it is numerically unstable and I should avoid it when I write program for QR decomposition
Householder reflections or Givens rotations are better choice for those who want to write a program
Silly , or maybe not in Householder reflections we need transpose of matrix to get Q and square matrices are easy to transpose in place
Transpose of rectangular matrix also can be done in place but it is not so easy
Thanks you for an amazing explanation!
thank you great video
Very useful session thank you so much ❤
WOW this was explained really well, I wish my professor could teach like this :(
A nice lin alg video again. I hope my students will also like it 😊
Thank you sir👍
Great video man
Thank you sir.
This was really good :)
thanks for your work.
very very clear
Do you have videos on SVD? Thanks for this video
crazy good, thank you
THANK YOU SO SO MUCH
The watch steals the show
3:16 what does he mean here? what is rescaling a vector ? just getting rid of the denominator ?
Multiplying by a constant, here so that the components are integers
Thanks this video series has really been helpful! Also, just wondering if you would be willing to share but I was wondering what watch you're wearing? I think it looks great
It’s an Invicta watch, you can get it on amazon
@@drpeyam Thanks for letting me know!
I wonder... Would a left-sided RQ decomposition ever be useful? And how easy is it to generate compared to QR?
good job peyam
Foi muito útil! Thank you so much!
What is the application of QR decomposition?
Hey, thanks for the amazing video. I just have one quick doubt. As you said Q is orthonormal matrix, then when I compute Q*Q' it does not give I. Please enlighten me, or am I misunderstanding something?
It doesn’t have to. For nonsquare orthogonal matrices we don’t always have Q Q’ = I, that’s only true for square matrices
Thanks! Sir
How about its importance in finding evalues?
thanks for the video! but one question, even if R' is not invertible (meaning A has linear dependency) , there still solution for LS right? just lose one dimension. No?
I thought u2 hat was perpendicular to V1, but apparently that's what v2 is?
Or is u2 hat supposed to be parallel to V1?
No, u2-u2hat is perpendicular to v1
What, if you define a kind of Matrix-Product with the compositon of the elements: A = { {a11(x), a12(x)},{a12(x), a22(x)} } and B = { {b11(x), b12(x)},{b12(x), b22(x)} }
So that, A°B = { { a11(b11(x)) + a21(b12(x)), a12(b11(x)) + a22(b12(x)) }, { a11(b21(x)) + a21(b22(x)), a11(b21(x)) + a21(b22(x)) } }
Linear Algebra is boring, because I never understood it well. Make more videos about this crazy fractional calculus stuff!
Something like this: (d/dx)^f(x) x = f'(x), where f(x) is the order of the derivative.
Or functions that transform other functios to their derivatives: g1(f(x)) = f'(x), g2(f(x))= f"(x), ... where gn depends of (d/dx)^n f.
Then you could "maybe" generalize derivatives by these functions --> g0.3(f) = (d/dx)^0.3 f
in W1 , where did you get 1/3 ? and what is W1? you said its the lenght of vector[2 2 1] it should be 3 . but why its 1/3?? i dont get it
You divide by the length of the vector to get a unit vector
@@drpeyam I just understand that numerator 1 is part of the formula right ? that was my question . 😄
thx bro !!
Great!
Good video! I looked very good. What do you think of Jimmy Hendrix? If you like my guitar and harmonica you will be happy.
Thaaaank yooooouuuu
IR=V fin.
Excuse my word choice, but this video is fucking amazing!!!
Thank you soooo much 😊
Dr Peyam.. can you please explain what LU decomposition is.. I kind of noticed in my textbook... but no idea what it is..
There’s a video on that
@@drpeyam Ok thanks... I'll check it out
Could you do QZ decomposition.
👏👏👏👏
Dr Ariya from Krish
6:50 known Q find R
Orthogonal Matrices are a bit strange anyway. I will never understand why you call a Matrix with orthoNORMAL column-vectors orthoGONAL and not orthonormal.
Most of the time I love maths, but sometimes I hate it xD.
I agree with you. Luckily, there is a book where authors say orthonormal matrix. In my lectures I also say this.
And there are other cases where I use the corresponding correct name. Similar, more terrible thing is 😄 "this infinite series is convergent". Math is nice, but the language created by persons is not necessarily correct.
@@sandorszabo2470 Finally someone understands me :D
I totally agree
Eh useless semantics
pleaseee in 03MIN.12 why the first resultat is 15/9 !!!
i love you
Do 100 integral challenge!!!
There’s already a 100 T/F challenge
08: 10
Your V's look exactly like your U's
t amo gringo, entendi como el putas no mk lo amo me ayudaste a estudiar para el parcial no nea feliz
「動画の音が良くない」、
Do you put the slash in the wrong place in "Q" just annoy viewers like me!?
No, that’s just the way I’m used to writing it
@@drpeyam fair enough
@@typo691 he is left handed so its easier to swipe down and away
First uwu!
I want to kiss him :*
Awwwwww!!!
The math people call this Q is semi-orthogonal matrix. They define that orthogonal matrix must be square.
en.wikipedia.org/wiki/Semi-orthogonal_matrix