I bought your book in 2018 from a random browsing at the local bookstore. Since then it has been the reference for my research. Thank you for making this course public as well!
This course is great! Big thank you! One quick question prof Roman: Are you planning to also make avaialable your other course about Brownian motion (Math 271A) ? Thank you again!
This is amazing. Thanks very much for these lectures. By the way, is that an 11' or 12.9' Ipad? I am hesitating between the two and my main usage will be taking notes similar to these. Many thanks
You are most welcome! This is iPad Pro 12.9 4-th generation, I bought it is 2020. The size makes a difference on the strain of my eyes, since I work on it several hours a day. So I think buying the larger one was well worth the difference in the cost.
For most of this course, dimension is the number of parameters of the data. More rigorously, we use the standard definition of dimension from linear algebra (see e.g. Wikipedia).
It still seems baffling that the 1/sqrt(n) bound will be independent of dimension. It is only natural that as we go in higher dimensions we will need more points for an accuracte estimate. What is the opposing force that keeps the number of points required for same accuracy a constant ? I seem to be missing the intuition for that. In other words, why is picking 5 random points in 1-d expected to give same accuracy as picking 5 points in 100-dimensions ? Thanks,
So the statment is about expected value of error. I think as we go in higher dimensions, the number of ways in which we can: 1. pick the "correct" points and end up getting a low error, and the number of ways in which we can 2. pick "wrong" points and end up getting even higher error increase proportionally. Is that right? Does the variance of the error increase with dimension or does it also stay constant?
I bought your book in 2018 from a random browsing at the local bookstore. Since then it has been the reference for my research. Thank you for making this course public as well!
Glad to hear it. Thank you!!
You can’t imagine how happy I was to see this open in TH-cam when I clicked the link on your site
Thanks for the kind words! I hope this course wil be useful for you.
@@romanvershynin2873 Hope you and your loved ones in Ukraine are all safe. Love and peace, I wish you all good health.
@@dtam4128 Thank you so much! All the best to you, too.
So happy to find Prof. Vershynin personally teaching us his book, so excited !
Great video! I'm so excited you released these. Your book is also incredible! Many thanks 🙏
You are most welcome, Stephen! It's great to know you find it useful, thanks for telling me this.
Dear Prof. Vershynin, your book arrived at my house today, and it's a beauty.
I am so grateful to have access also to your lectures, thanks!!!
You are most welcome! I am glad that my you enjoyed my book and this video course.
Like your book! We have a study group to read the book! glad now we have the official courses!!!
Glad to hear that you are enjoying studying my book in a group setting. Have fun!
Excellent resource, thanks a lot for making this public!
You are most welcome!
Thank you so much for making this public
You are most welcome!
Thanks so much. Really am appreciative of your course and book.
Greetings from Ukraine. I found your course by chance and I am very happy with it, thank you for this opportunity)
Happy to hear that you enjoyed my course! My best wishes to you in Ukraine.
Appreciate providing the lecture on youtube!
Much help to me! Love from China.❤
Glad to hear it! Thank you. Cheers from Califiornia.
Really appreciate the lectures & the book 🙌🙌
very nice, I eas looking for such a course for a long time.
my professor!!!!!!👍👍👍👍👍👍👍
Thank you so much!
This course is great! Big thank you! One quick question prof Roman: Are you planning to also make avaialable your other course about Brownian motion (Math 271A) ? Thank you again!
Thanks, Julio! I am not planning to publish 271A at this time.
This is amazing. Thanks very much for these lectures. By the way, is that an 11' or 12.9' Ipad? I am hesitating between the two and my main usage will be taking notes similar to these. Many thanks
You are most welcome! This is iPad Pro 12.9 4-th generation, I bought it is 2020. The size makes a difference on the strain of my eyes, since I work on it several hours a day. So I think buying the larger one was well worth the difference in the cost.
Thank you for this great course.
You are most welcome!
Thanks
Thank you so much! I really enjoyed this class💯💯💯
I am happy to hear this! You are most welcome.
Thank you so much!
Could you give a definition for dimension in this course?
For most of this course, dimension is the number of parameters of the data. More rigorously, we use the standard definition of dimension from linear algebra (see e.g. Wikipedia).
It still seems baffling that the 1/sqrt(n) bound will be independent of dimension. It is only natural that as we go in higher dimensions we will need more points for an accuracte estimate. What is the opposing force that keeps the number of points required for same accuracy a constant ? I seem to be missing the intuition for that.
In other words, why is picking 5 random points in 1-d expected to give same accuracy as picking 5 points in 100-dimensions ?
Thanks,
So the statment is about expected value of error.
I think as we go in higher dimensions, the number of ways in which we can:
1. pick the "correct" points and end up getting a low error, and the number of ways in which we can
2. pick "wrong" points and end up getting even higher error
increase proportionally. Is that right?
Does the variance of the error increase with dimension or does it also stay constant?
Nice shirt! ;)