Lecture 15 | Lagrange Dual Problem | Convex Optimization by Dr. Ahmad Bazzi

[@cutestbear3327]

Thanks for the smiley face Dr. Bazzi, your detailed, patient way of explaining and untangling complex abstract concepts does indeed put a smile on my face :D

[@justinjose4361]

Sir you teach very well. I implore you to complete the lecture series by adding more videos frequently.

[@AhmadBazzi]

thank you for your words. Indeed, I will keep uploading on the series. You are also invited to follow my Convex Application series, which is more "fun" as it supplies real life examples that could be solved by convex optimization.
ruclips.net/video/38u85fHxU-M/видео.html

[@thomasjefferson6225]

​@Ahmad Bazzi maybe this summer. I'm in a math course, not an economics one!😅😅

[@Aarshyboy96]

This is great. Thank you for sharing your lectures!

[@AhmadBazzi]

Glad it was helpful!

[@nguyenluu3082]

I am interesting this course even i've just started on it... Could you plz let me know how do we can deeply understanding the theory to solve the problem? I have read the convex optimziation book, and cannot understand all !
And Thanks for your great lectures, Dr Ahmad!

[@santhuathidi5987]

sir please make a video on support vector machine

[@YuxuanHarry]

Hello, Bazzi! At the Ex.8, shouldn't we consider the constraint \lambda >= 0? Thx!

[@jeanishimwe2120]

When to derive the dual function or to use the supremum?

[@davidleedavid]

at 57:54 I understood x_{0} = inv(P(\lambda)) q(\lambda). How does x_{0}.T become q.T(\lambda) inv(P(\lambda) rather than q.T(\lambda) inv(P.T(\lambda)) ?

[@AhmadBazzi]

But P(\lambda) is symmetric, so P^T = P

[@iamdrfly]

at 27:16 the alpha of the convex hull should be greater equal to zero, am i wrong ?

[@felixhue9254]

You are a great teacher, but you are too fast and you have too much advertisements added. ^^

[@bakerf2431]

at 47:23, u^Tc < 0 leads to the conclusion of -u^Tc < 0??

[@AhmadBazzi]

Not at all, u^T c < 0 means - u^T c > 0

[@bakerf2431]

Ahmad Bazzi exactly, I’m wondering how the derivation on the lower right corner get to the conclusion that 0>-u^Tc>u^TA^Tz

[@AhmadBazzi]

@@bakerf2431 -u^Tc>u^TA^Tz comes from the separating Hyperplane theorem

[@nickey0207]

@@AhmadBazzi Hi Dr Bazzi, from the above, 0>-u^Tc>u^TA^Tz and - u^Tc > 0, I suspect the term 0>-u^Tc>u^TA^Tz shall be written instead as -u^Tc>u^TA^Tz>=0? Please correct me if i am wrong. Cheers

[@reginaochonu-ocheja6064]

A question on the Ellipsoid Algorithm for convex optimisation
Hello Dr. Ahmad,
I am currently following these lectures on Convex Optimisation and it's been great, especially since it's my first time going into convex optimisations. So, THANK YOU SO MUCH!
I have to use the Ellipsoid Algorithm to find the lagrangian (lambda) in a dual problem. But I am finding it difficult to apply the algorithm to the problem in question (found in equations 10 through 17 in the paper: arxiv.org/pdf/1911.03264.pdf )
Do you think you can help?
Thank you so much in advance.