Matrices of Chocolates (and Delicious Missing Data)

Hmmm... this seems like a great way to demonstrate linear algebra and matrices. And then to demonstrate matrices with progressively more missing data... (Delicious missing data.)

A rectangular Hankel chocolate box. You need O( log( m + n ) ) time to eat it. pic.twitter.com/jkhYQSb9mT

— Daniel Kressner (@_kreda_) March 20, 2019

A Cartoon Depiction of Deep Learning Versus Traditional Machine Learning

So true. :)

great meme pic.twitter.com/d5AW6WliTy

— Lex Flagel (@flagelbagel) July 29, 2018

(Tip of the cap to Michael Stumpf.)

Almost 11 Years in Review

11 years passed between this paper's initial submission and its final acceptance.

no word needed... @YourPaperSucks pic.twitter.com/WW3zue9kmO

— Sarah Lambart (@Sarah_Lambart) December 8, 2017

For the story, see this Quora answer

(Tip of the cap to Joel Miller.)

A Big Pile of ... Linear Algebra

That's right. In many cases, machine learning is a big pile of ... linear algebra.

NFL Offensive Lineman by Day, Spectral Graph Theorist by Night

John Urschel is not your ordinary professional football player.

He may be an NFL offensive lineman by day, but by night he is a spectral graph theorist (and numerical linear algebraist). His most recent paper, called "A Cascadic Multigrid Algorithm for Computing the Fiedler Vector of Graph Laplacians", has now been accepted for publication in Journal of Computational Mathematics. Urschel announced via Twitter that it had been officially accepted for publication. (Based on my googling, the published version of the paper hasn't yet appeared in the journal.)

You can read a draft of Urschel's paper on the arXiv preprint server. I just wish that he used his current affiliation on the paper. That would have been fantastic.

(Tip of the cap to Francis Su.)

Update: Looking at Urschel's academic website, I see that he has prior publications. His website gives the reference for a paper on celestial mechanics. I also checked Mathematical Reviews (to find an upper bound on his Erdős Number, of course), and I see that he also has at least one more paper on spectral graph theory. (This paper isn't listed on Urschel's Penn State website, which I suppose he is no longer updating.)

Update (3/21/15): I wrote a blurb on Urschel for the Improbable Research blog.

Upset Triangular Matrices

I just saw a delightful typo in an e-mail that I received a few minutes ago: Apparently, there is something called an "upset triangular matrix". I better go and console it.

I think that that typo just made my day.

Headline: Sudden Execution Causes Extinction of Matrix Products

Here is a quote from a paper I finished reading this morning: "We first execute the last matrix product..."

Well, damn. Now I guess matrix products have become extinct. Cruel bastards.

The Basis of a Good Outfield

An article tagline that I would love to see in a newspaper: "Jayson Werth, Bryce Harper, and Denard Span form the basis of a solid Nationals outfield."

Sadly, I doubt most newspapers like linear-algebraic jokes as much as I do. Maybe an independent newspaper?

Justice Scalia learns about orthogonality.

This excerpt from a recent U.S. Supreme Court deliberations transcript is pretty damn funny.

(Tip of the cap to the 'Found Math' column in MAA Focus. This was in the Feb/March 2010 issue.)