Gilbert Strang's final lecture at MIT: May 15, 11:00am | Hacker News Reader

Gilbert Strang's final lecture at MIT: May 15, 11:00am

grinfeld.org

263 pointsdeepzn56 comments

Linear Algebra was a course in the first semester of freshman year when I was in uni (this wasn't in the US or in an english speaking country). It was famous for being a "filter" course, that made a lot of people quit the degree and switch to a different degree next year.

It had a 94% failure rate, but I got to pass on the first attempt because Gil's free lectures of his LA course made me intuitively understand the stuff we were being taught in that course, which was very dry and IMO poorly explained, at least for freshmen. Don't think I would have dodged the bullet otherwise.

Thanks Gil.

I had the same experience. I didn't understand the teaching from my university. Once I found his lectures it all clicked. I understood what we were doing. It was no longer wrote learning of algorithms to manipulate things with no meaning.

I did very well in that paper.

Fwiw the word is “rote” not “wrote”

TIL, thank you.

I <3 HN

:))

I have a feeling that the intellect gap at a lot of universities isn't as profound as one would expect and that one of the reasons that the students of the best unis do so well is because their educators can actually explain the course material. You can do more when you've got to spend less cognitive energy on trying to piece together the hieroglyphic puzzle that lesser professors throw at you because they believe they're too important to teach.

A lot of the top universities have majority of their profs caring almost exclusively about research though. I think MIT is (or at least was) a bit of an exception, where there was enough of a critical mass of profs that cared about teaching. Can't say the same for a couple other "top" unis I've spent time at.

Yes I've heard that as well. For example I heard 2nd hand that Andrew Wiles at Princeton, who famously solved Fermat's last theorem, is a brilliant researcher but a terrible professor.

I had a math professor from Dartmouth who told me that Dartmouth really emphasized teaching in their phd program and he was indeed an exceptional teacher. I wish that was emphasized more, or they separate hardcore researchers from the teachers.

It’s a good point, although Princeton in general places a higher emphasis on undergraduate teaching than a number of other top research universities, so that example might be idiosyncratic if true.

The emphasis on teaching in a PhD program may not have much bearing on the quality of undergraduate teaching at a university, though it might be indicative of a culture or spirit.

I would have to agree with this. In all my time in school I've had terrible and amazing professors, the difference in the amount I learned was generally due to their ability to explain the subject in a way that was easy to understand. I also credit the MIT open courseware chemistry lectures for helping me through uni, because that professor was one of the not so great ones at my school

i find that in general (not always) the smarter people are, the better they understand precisely what you don't know, and what you need to know to understand what they are trying to tell you

it's a bit like the old "the more I know, the more I know I don't know" but with an addition of "the more I know exactly what I know, and exactly what I don't" and it's then applied via theory-of-mind, "the more I know precisely what I can teach you"

one related quote i really like is : "it takes a lot of knowledge to realize the extent of your own ignorance" :D

I agree with u on Strang, wonderful lectures. I always learned on books at least a decade or older, I was amazed even the 90s about his lasting legacy. He has been helping people with linear algebra for more than 50 years. I haven’t even been alive for that long.

Weeder courses shouldn’t exist, they are immoral. Glad you survived yours.

Such an interesting principal to carry, I never thought about it before.

Now that I am, it seems kinder to fail people early out of paths they are ill suited to follow, no?

If it’s something they’re interested in or need to know for their career, then I don’t think it’s possible to be ill-suited, only ill-prepared.

It’s mostly because in free public universities there’s an appetite for making things hard. Very hard. The less students graduate, the most prestigious they think they are

I took linear algebra about 20 years ago as a requirement for my CS degree. I had no clue whatsoever what it was for beyond just "solving systems of equations" - whatever that meant. The prof didn't impress upon us how it underpins all of machine learning (which, mind you, wasn't as much of a hot topic then that it is now). Our attempts at grokking its importance were largely met with a kind of "because it's important" dismissal.

I didn't do great in the course, and when it was done, I was pretty happy this useless subject was behind me.

It wasn't until several years into my profession that I learned how important linear algebra truly is. I now work very closely with tons of data scientists, and I still feel like I should go back and learn it better.

Same thing happened to me. Yes ml wasn't a hot area then (20 years ago) but computer graphics, linear regression (ie old school statistics used for supply chain), signal processing for audio and video were all got them but their applicability was never pointed out let alone emphasized. I was just lucky that I was interested in these areas on my own so the linkages thankfully "clicked". Is it me or did profs back then just do research and teaching in a bubble?

Multiple generations of scientists and engineers are forever indebted to Strang: He's been teaching for over six decades.

At first only MIT students were lucky to learn from him, but then, as his videos became available online, so did the rest of world.

I've yet to meet anyone working in AI, ML, data science, signal processing, etc. who hasn't watched at least some of Strang's lectures.

I've also met multiple people working in these/related fields for whom Linear Algebra 'clicked' only after watching Strang's lectures.

He deserves the collective applause and giant thank-yous he's getting.

I hope MIT names something after him.

I had the privilege of taking one of Strang's classes a few years back. He is still incredibly sharp, and has a great sense of wit.

During one lecture, he was half way through solving a problem on the board when he jumped straight to the final answer. He then turned to us, and with a deadpan delivery said:

"We haven’t proved it, but that’s okay, we only live so long."

And then he moved on to the next problem.

I've quoted that line a lot since then.

Here are 34 lectures for his Linear Algebra course he recorded in 1999:

https://ocw.mit.edu/courses/18-06-linear-algebra-spring-2010...

Along with Jacques Pepin's "deboning a chicken" .FLV video, these are among the greatest videos on the Internet.

Wow! I'm not sure how I missed that back in the day.

Here's the video transcoded on YouTube for anybody else who's curious:

https://www.youtube.com/watch?v=nfY0lrdXar8

Great comparison! Both are absolute treasures.

IDR which lecture it was but there's a timely Monica Lewinsky joke somewhere in here. Small things but I appreciate when profs are allowed to have fun. Obviously Strang is also just an amazing teacher.

I picked up his linear algebra book and watched his lectures last year as part of my journey into machine learning. He made me fall in love with math in a way that I never had before, not even during my physics degree. Truly an inspiring teacher and amazing person.

Agree. I’m seriously thinking of starting a GradCert in October thanks to his teaching.

Strang's bio: https://en.wikipedia.org/wiki/Gilbert_Strang

He's been a professor at MIT since 1962!

Not to steal limemlight away from Strang, but I'd recommend again Pavel Grinfeld's (the website host, also Strang's student (I think)) Linear Algebra series: https://www.youtube.com/playlist?list=PLlXfTHzgMRUKXD88IdzS1....

I still have his linear algebra book on my shelf from the 1980's. One of the great teachers of a topic that has been crucial to my career in computer graphics and vision.

Fully agreed. I would argue that there is no replacement for Strang, though many may try. My copy has only been around for ~20 years, well worn from when I struggled with the course.

Very sad to see him go. I am fully envious of those who will be in attendance.

I took his applied linear algebra course as an undergrad. Such a kind man and so passionate about teaching. Very much "in the know" on modern ML, like LMs, and he found ways to tie the fundamentals to current topics. Always happy to sign a copy of a his books :)

Cheers to his post-Institute era.

Thanks Prof Strang.

Are there any "side channels" that host university lecture videos and course materials for less fortunate students in 3rd world countries? The ones on MIT courseware or Youtube are old.

I mean something along the lines of academic torrents etc...

The good news about 18.06 is that it doesn't need to be new, just good, and it is that already.

Huh? There's new university lecture videos being added to Youtube all the time, so I don't know why you'd say everything there is old. Besides, for many (most?) subjects it doesn't matter if the videos are old or not. Something like, eg Linear Algebra, just doesn't change that much.

All of that said, one other option you might check out is videolectures.net[1]. There's some pretty good stuff there.

[1]: http://videolectures.net/

In the late 80s I bounced off of a couple of linear algebra courses (when MIT had a lot of trouble teaching undergrad courses in advanced topics because they assumed it was just prep for the graduate course instead of actual teaching.) Then I got in to Strang's course, and it was such a breath of fresh air - he was honestly, effusively enthusiastic about the subject and this was contagious. "Engaging" doesn't do it justice :-) He put a lot of work into both demystifying and clarifying jargon, and explicitly "connecting the dots" between related concepts.

There is a great book published in Russian "ЗАДАЧИ И ТЕОРЕМЫ ЛИНЕЙНОЙ АЛГЕБРЫ" (Problems and theorems in linear algebra), by Victor V. Prasolov. English translation was done by American Mathematical Society. Unfortunately, this translation is for the first edition of the book. Latest edition is available for free: http://prasolov.loegria.net/linalg.pdf (hint is you want to print it out, Amazon printing service is great place to print books)

I think this is not actually the newest edition; that is available on https://sites.google.com/site/prasolovskacatmoiknigi/home .

It's a great book, though for a completely different audience than Strang. It collects hundreds of apocryphal results from 400 years of linear algebra, most with proofs, some fairly deep.

Here's the English 1st edition: https://staff.math.su.se/mleites/books/prasolov-1994-problem...

Amazon printing service? I see they have one for photos/calendar and such. Do they have one for sending PDF's to print?

omg https://kdp.amazon.com/en_US/help/topic/G201834340

ah crap no pdf

I think, that is the Amazon service I have used to print out the book. I uploaded the pdf and week later got the book from Amazon. I will go though email and will get back to you.

Yep, it is Amazon service. Here is the email that I got from Amazon. Make sure you order book as Request copy.

Dear , Your request for a proof version of ‘Задачи и теоремы линейной алгебры’ is ready for checkout. Requested copies will be available in your Amazon shopping cart for the next 24 hours. You can complete checkout by clicking the following link:

Proceed to Checkout

Questions? Learn more on our help page.

The Kindle Direct Publishing Team

I used his book on linear algebra during my studies in Germany, even though the professor teaching LA pursued a different (and I found more confusing) approach. I found the book really enjoyable, in my mental bookshelf it rests right next to the Feynman lectures in physics.

I took a course from him 49 years ago.

Thanks to his recorded lectures I finally passed LA in a dutch University. At the time ~2008 my university profs had a say for specific courses if they wanted it recorded or not. Of course the profs who refused were gray dinosaurs. I'm glad Gilbert didn't fall into this education conservatism.

I taught a linear algebra section for Gil when I was a postdoc, will be watching this for sure.

Ha, I knew I recognized that name in the domain. Pavel Grinfeld has some great lectures on YouTube, and last I looked, was working on a very useful interactive site for learning mathematics.

This is his channel:

https://www.youtube.com/channel/UCr22xikWUK2yUW4YxOKXclQ

I love it, especially the videos on Linear Algebra and Differential Equations. He's an extraordinary teacher, much like Strang, imo.

Oh so this is just a retire in peace .. I was affraid for a minute.

You thought he’d slipped into the null space, but instead he found his dominant eigenvector.

comments like these keep make me coming back to hn :) witty comments ftw !

That's Gil for Gilbert, not GIL for, idk, Python's Global Interpretor Lock or something.

Because this has suddenly gone from +something to -4 I want to point out that when I wrote it, the submitted title was 'GIL Strang [...]'; so it made a lot more sense (and the title didn't)...

It'd be nice if a title change afforded a possibility to edit/delete comments, or if mods doing so would scan the comments for anything about them.

(I recently had a very old tweet liked and retweeted by someone I mentioned in it, that I can't make head or tails of - I can only assume the account is now owned by someone else - so I'm a bit sensitive to it...)

Yes, thanks edited it after seeing your comment.

I really enjoyed using his first textbook. He's got a nice, light touch.

I wonder if enough people will tune in to a linear algebra lecture that it crashes the live stream.

Reply on news.ycombinator.com