Show more

New blog post: Circles crossing at equal angles,, proving that if a non-self-crossing quadrilateral with circular-arc sides has four equal interior angles at its vertices, then it is cyclic.

UCLA suggests that its faculty refrain from publishing with or reviewing for Elsevier while negotiations are ongoing:

For those willing to take a longer-term stand, there's always

This Sudoku is (as always) easy with backtracking, but it seems harder than most of the ones that are hard enough for my program to need backtracking:

2 * * * * 9 * 1 *
* * * * * 3 * * *
* 3 5 * * * 2 * *

* * 4 * * * * * *
* 8 * 6 * * 3 9 *
* * * 8 2 * 7 * 4

* * 8 7 * * 4 * *
* 9 * * * 6 * 3 *
* * 1 * * * * * *

Choose a polynomial's coefficients randomly and independently from your favorite nontrivial distribution. Then it should be irreducible with high probability for polynomials of high enough degree. This was previously conjectured for the uniform distribution on \(\{0,1\}\) by Odlyzko and Poonen; now Breuillard and Varjú have proven that it follows from a form of the Riemann hypothesis. See: (preprint at

Electric eel inspires biocompatible hydrogel battery:
(based on paper at; see also

The part that caught my attention is that they're using a Miura fold to simultaneously align and press together many pairs of droplets of four types (salty, fresh water, or two kinds of charge-selective hydrogel), creating an origami-activated electrical discharge.

Byrne’s Euclid recreated for the web:

Beautiful three-color figures, hard-to-read old-faſhioned orthography, and all. I have the Taſchen reprint in my office, but I prefer the Dover Heath edition for actually uſing the books rather than looking pretty.

Via and

LaTeX, the game:

Via @ejk,

How many extra levels for getting the commutative diagram to format in Wikipedia's lobotomized version of LaTeX?

A move to double-blind reviewing fixes huge gender bias in acceptance rates for space telescope allocation competition:

xkcd on arXiv vs the commercial journals:

(Infer what you will from the fact that xkcd is published open-access.)

Map of the tracks of six wolf packs shows them carefully partitioning the wilderness into non-overlapping territories, and staying within their own:

Higher resolution version at the linked reddit page (probably also at the linked facebook page for anyone who actually cares to visit facebook).

Flip graphs of matroids are expanders:
Gil Kalai reports on a new proof by Nima Anari, Kuikui Liu, Shayan Oveis Gharan, and Cynthia Vinzant of a conjecture by Milena Mihail and Umesh Vazirani that any subset \( S \) of at most half of the bases of a matroid has at least \( |S| \) flips to a base outside \( S \).

Based on a preprint at, via

Taylor and Francis doing Trump’s dirty work for him:

Timothy Gowers reports on the journal _Dynamical Systems_ refusing to publish a paper (supposedly, after editorial acceptance because one of its coauthors is Iranian.

New blog post: General-position hypercube projections

In which I look for large general-position subsets of 2d grids (the no-three-in-line problem) by starting with higher-dimensional point sets and then carefully flattening them. So far the results are only computational but they look promising.

Australia's war on encryption: the sweeping new powers rushed into law

With all the vague "end of the internet!" or "why is everyone so upset at such good idea" stories elsewhere on the new Australian backdoor-your-apps law, it's good to see a piece that explains what it actually does.

The moral seems to be: don't allow or agree to any automatic updates while in Australia. Australians, if you want security from govt snooping and from security flaws, tough luck.

ACM announces their 2018 class of fellows:
Congratulations, new fellows!

You can also see Wikipedia's coverage of all the ACM fellows at
Quite a few are still missing articles, so if you want to improve Wikipedia's coverage of computer scientists there's still plenty to do.

Rigid Foldability is NP-Hard:

It was previously known that folding a purported origami folding pattern to a flat state is NP-hard, because you can encode logic in the way the paper gets in the way of itself. But this paper proves that it's hard even to tell whether you can make any rigid motion at all starting from completely unfolded paper, well before self-interference kicks in. Instead, the difficulty involves getting sums of angles to come out right.

Show more

A Mastodon instance for maths people. The kind of people who make \(\pi z^2 \times a\) jokes.

Use \( and \) for inline LaTeX, and \[ and \] for display mode.