Recently added

What's new

Essays arrive in groups rather than one at a time. The most recent group is below in full, and every earlier one after it, newest first.

Essays arrive in groups rather than one at a time, and a group usually opens up a subject not covered before. Between one group and the next nothing changes, so a reader who has seen the most recent group has seen everything.

12 November 2026

22 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

Two sign sequences laid on themselves. Length 41: random signs (seed 41) with E = 552 and merit factor 1.523; the rotated Legendre sequence of 41 with E = 176 and merit factor 4.776. Algebra

A quarter turn quiets the echoes

Lay a sequence of signs on a shifted copy of itself and add up the squares of the overlaps. Random signs leave the total near n²/2, and the squares modulo a prime, started a quarter of the way along, cut it to a twelfth. Exhaustive search finds far quieter short sequences, and no infinite family is proved to do better than 6.342.

6 figures
Coefficients that close in on a positive curve. The form x² − 1.8xy + y² on the segment x + y = 1, minimum 0.050, with the coefficients of (x + y)ᴺ times it, divided by binomial coefficients, for N = 0, 6 and 19; at N = 19 all are positive. Algebra

Multiply until every sign is positive

A form with a negative coefficient can still be positive wherever its variables are, and proving it usually takes an argument. Pólya found a certificate that needs none: multiply by a power of x + y and wait until every coefficient is positive. The power needed is finite exactly when the form never touches zero, and it grows like the reciprocal of how close it comes.

6 figures
Rebuilding a ball at its own centre. Fourier inversion of the indicator of the unit segment, disc and ball, keeping frequencies below R, evaluated at the centre: at R = 8 the values are 0.9873, 0.9206, 0.9873; in three dimensions at R = 7.75 it is 1.6369. Analysis

A ball rebuilt everywhere but its centre

Rebuild a solid ball from its frequencies, keeping those below a cut-off, and read the result at the centre, where the ball is perfectly smooth. In the plane the answer settles on 1. In space it swings between 0.36 and 1.64 for ever, because the whole surface of the ball is the same distance from the centre and its jump arrives there in step.

6 figures
The curve whose bending grows with its length. Euler's spiral, curvature equal to arc length, for s from −7 to 7, winding into ±(0.88623, 0.88623). Analysis

The curve that bends as far as it has gone

Say how sharply a curve bends at each distance along it, and the curve is fixed. Bending in proportion to distance gives Euler's spiral, whose arms are infinitely long and end a finite distance away. Bending that repeats decides at once whether the curve ever closes, and taken in steps the spiral is a Gauss sum.

6 figures
Three widows and the Mishna's table. The Mishna's division of estates of 100, 200 and 300 among claims of 100, 200 and 300: 33⅓ each; 50, 75, 75; 50, 100, 150. Applied

The debts the Mishna divided

A man dies owing his three wives 100, 200 and 300. The Mishna divides an estate of 100 equally, of 300 in proportion, and of 200 as 50, 75 and 75, and for nearly two thousand years nobody could say why. The table is the nucleolus of a cooperative game, and the only division in which every pair splits its share as two people split a contested garment.

6 figures
Six goods, three people, and no envy. Three people's random values for six goods, with an envy-free allocation shaded: Ada values the bundles at 0.86, 0.85, 0.34; Ben values the bundles at 0.64, 1.83, 0.75; Cy values the bundles at 1.05, 1.24, 1.29. Applied

Twice as many goods as people

Indivisible goods with random values cannot always be shared so that nobody envies anybody. With one good each it works only when everyone's favourite is different; with twice as many goods as people it almost always works. Deciding each case exactly shows the chance climbing steeply in between — and a simple draft getting there only when the goods divide evenly.

6 figures
Every convex shape the tangram can make. The 13 convex polygons that the seven tangram pieces can form, each with one arrangement, and the number of arrangements of each: 20, 12, 12, 12, 36, 12, 8, 8, 16, 4, 56, 24, 36. Computation

Thirteen convex tangrams

The seven tangram pieces make thousands of figures, and only thirteen of them are convex. The proof turns a puzzle with sliding pieces into a finite list: every piece is made of the same small triangle, any convex figure puts them all on one grid, and that grid allows exactly twenty convex outlines of the right area. A search tries all twenty and finds that thirteen can be made.

6 figures
Seven hats, sixteen losses. All 128 ways seven red-or-blue hats can fall, each drawn as seven dots, with the 16 configurations the Hamming strategy loses framed: the team wins 112 of 128, a probability of 7/8. Computation

Wrong guesses that arrive together

Seven players in red or blue hats each see every hat but their own, and must guess their own colour or pass. Every guess is a coin toss, yet with the Hamming code the team wins seven times in eight — not by making guesses better, but by making the wrong ones arrive together. The best chance for any number of players is fixed by the fewest losses that cover the cube.

6 figures
Three edges at every point, and how many ways to pair. The 3-regular bipartite graphs with six vertices a side having the fewest (17) and the most (36) perfect matchings among all 297200, each with one matching in red. Discrete

How many ways a regular graph pairs off

Give every point of a bipartite graph the same number of edges and Hall's theorem promises a perfect matching. It promises far more: at least ((d − 1)^(d−1)/d^(d−2))ⁿ of them, by Schrijver's theorem, and at most (d!)^(n/d), by Brégman's. Listing all 297,200 cubic graphs on six and six finds only four different counts, and random graphs turn out to live near the bottom.

6 figures
Fifteen edges that need four colours. The Petersen graph with its 15 edges in 4 colours; a search over all colourings with three finds none. Discrete

One colour more than the busiest point

Colour the edges of a graph so that edges meeting at a point differ. The busiest point forces as many colours as it has edges, and Vizing proved that one more is always enough. So every graph is in one of two classes, and deciding which is hard — yet almost every graph is in the first, and the ones in the second carry an odd piece that cannot be paired off.

6 figures
Two blocks and a wall count to 31. Space-time diagram of a heavy block of mass 100 driving a block of mass 1 against a wall: 31 elastic collisions in all. Dynamics

Two blocks and a wall count π

Slide a heavy block towards a light one resting by a wall, and count every collision until they part for good. A heavy block 100 times the mass gives 31; ten thousand times, 314; a million, 3,141. The digits are π because the two blocks are secretly one billiard ball in a wedge, and a ball in a wedge bounces as many times as the wedge's angle fits into half a turn.

6 figures
Islands of order in a chaotic sea. Poincaré section of the Hénon–Heiles system at energy 1/8: orbits started on a line of the section, crossings coloured by a separation test, 16 regular and 4 chaotic. Dynamics

A conserved quantity that is not there

In 1964 Michel Hénon and Carl Heiles asked whether a star in a galaxy obeys a third conservation law besides energy and angular momentum. They cut its orbit by a plane and looked. At low energy every orbit drew a clean curve, as if the law held; at higher energy the curves dissolved into dust. Integrating their model again shows the share of chaotic orbits climbing from nothing at energy 1/12 to almost all at 1/6.

6 figures
Four perpendiculars from inside, two from outside. Normals to the ellipse x²/4 + y²/1.44 = 1 from (0.55, 0.18) inside its evolute (4) and from (2.3, 0.9) outside it (2), with the evolute drawn. Geometry

Four perpendiculars to an ellipse

From a point near the centre of an ellipse, four lines meet the curve at right angles; from a point far away, only two. Apollonius found the boundary between the cases twenty-two centuries ago. It is the curve traced by the ellipse's centres of curvature, a four-cusped star, and it is also the envelope of all the normals, the fold in a family of lines where two perpendiculars merge and vanish.

6 figures
Four rods, hinged, enclosing the most. Quadrilaterals with sides 3, 4, 5, 6 as one corner angle varies; the largest has area 18.97367, equal to Brahmagupta's √360 = 18.97367, with its corners on a circle of radius 3.2873. Geometry

The most four hinged rods can enclose

Four rods of fixed lengths, hinged at their ends, can be pushed into a whole family of quadrilaterals. The one enclosing the most area has its corners on a circle, and Brahmagupta gave its area in 628 as a product of four terms. Order the rods any way round the circle and the area is the same; add a fifth rod and the circle still wins, but the formula is gone.

6 figures
Five ellipses, every combination once. A rotationally symmetric Venn diagram of five congruent ellipses (centre distance 0.7, semi-axes 1.4 and 0.5, tilted 7.5°): 32 regions on a raster, 30 crossing points. Logic

Venn diagrams that turn into themselves

A Venn diagram shows every combination of in and out exactly once. Ask for one that looks the same after a turn of 1/n, each curve rotating onto the next, and a counting argument forbids it unless n is prime. Five congruent ellipses manage it, certified by Euler's formula; four rotated copies cannot, and it took until 2004 to show that every prime can.

6 figures
Excluded middle fails at one point, and its double negation does not. Open sets of the real line as truth values: p = (0, ∞), ¬p = (−∞, 0), p ∨ ¬p misses 0, ¬(p ∨ ¬p) is empty and ¬¬(p ∨ ¬p) is the whole line. Logic

Every tautology survives double negation

Intuitionistic logic refuses excluded middle, Peirce's law and every other classical principle that claims more than can be shown. Glivenko proved in 1929 that it refuses nothing outright: put ¬¬ in front of any classical tautology and the result is intuitionistically valid. Every Kripke model of up to four states confirms it for eight famous laws, a census of 11,451 formulas confirms it one by one, and a model with a growing domain shows where it stops.

6 figures
Five powers of two that are seven more than a square. How close 2ⁿ − 7 comes to a square for n from 3 to 40, as a fraction of the gap between neighbouring squares: squares at n = 3, 4, 5, 7, 15 (1, 3, 5, 11, 181 squared), and at no other n up to 10000. Number

Five powers of two, seven more than a square

2³ − 7 = 1², 2⁴ − 7 = 3², 2⁵ − 7 = 5², 2⁷ − 7 = 11², 2¹⁵ − 7 = 181². Ramanujan asked in 1913 whether there are any more. Testing every exponent up to 10,000 finds none, and Nagell proved in 1948 that none exist, by moving to a number system where 2 splits into two primes and factorisation is still unique. The same five solutions decide when a two-error-correcting code can be perfect.

6 figures
Every way to make change for two pounds. Number of ways to make each amount up to 200 pence from British coins, on a logarithmic scale: 451 for 50p, 4563 for £1, 73682 for £2; a dollar in United States coins has 292. Number

Two pounds in 73,682 ways

Count the ways to pay an amount exactly with a fixed set of coins and the answer is almost a polynomial in the amount — a polynomial plus waves that repeat with the periods of the coins. With parts 1, 2 and 3 the whole formula fits on a line; with British coins the leading term is still 3% off at four hundred pounds. And the fewest coins is a different question, which the cashier's rule answers correctly only for some coinages.

6 figures
Where to look after looking and not finding. A prior over a 16 by 12 grid and the posterior after 60 looks by the greedy rule that found nothing: the chance of having found the object by then was 0.425. Probability

Where to look after looking and not finding

A lost object is somewhere on a map of probabilities, and every search that fails redraws the map. Bayes' rule says how: the cell searched loses probability in proportion to how good the search was, and every other cell gains. Looking where the chance of finding it on the next look is highest is provably best, and a search that overrates its own eyesight looks in the wrong place — which is how the wreck of Air France 447 was missed for two years.

6 figures
One swap splits the long cycle. A shuffle of 100 boxes whose longest cycle has 61 boxes, drawn as a ring; swapping the contents of two boxes 30 steps apart splits it into cycles of 30 and 31, so every cycle has at most 50 and all prisoners succeed. Probability

One swap saves every prisoner

A hundred prisoners following the cycles of a shuffle all survive about 31% of the time. Let a helper look into the room first and swap the contents of two boxes, and they survive every time. With fewer boxes each, more swaps are needed — exactly one per cut a long cycle needs — and the number of swaps a random room demands has an exact distribution.

6 figures
A knot written as a fraction. The two-bridge knot K(17/5) drawn as a plat on four strings from the continued fraction 17/5 = 3 + 1/(2 + 1/2), with 7 crossings. Topology

A tangle that is a fraction

Two ropes, four people, two moves: twist the right-hand pair, or turn everyone a quarter of the way round. Conway noticed that every tangle made this way carries a fraction, that the moves act on it as x + 1 and −1/x, and that the fraction is a complete invariant: two such tangles are the same exactly when their fractions are. Close the tangle and the fraction names a knot.

6 figures
One loop that outlasts the noise. 44 noisy points round a circle joined at scale 1.09, and the scales at which each loop appears and is filled: the longest from 0.588 to 1.588, the next 0.025 long. Topology

A hole that lasts across scales

A cloud of points sampled from a circle has no hole at all, and joined at the right scale it has exactly one — but nobody knows the right scale in advance. Persistent homology joins the points at every scale at once and records how long each hole survives. The circle's hole lasts; accidental holes vanish almost at once; and moving every point by δ moves every survival time by at most 2δ.

6 figures

Before that

Everything published earlier, newest first. Titles only — the cards are on the full listing.

5 November 2026

22 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

2 November 2026

22 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

11 October 2026

44 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

7 October 2026

44 essays on algebra, analysis, applied, discrete, dynamics, geometry, computation, logic, number, probability and topology

1 October 2026

11 essays on algebra, analysis, applied, discrete, dynamics, logic, geometry, probability, number and topology

29 September 2026

65 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

22 September 2026

22 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

16 September 2026

44 essays on discrete, applied, analysis, number, probability, topology, geometry, algebra, computation, dynamics and logic

13 September 2026

22 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

9 September 2026

33 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

8 September 2026

21 essays on algebra, analysis, applied, computation, discrete, dynamics, geometry, logic, number, probability and topology

5 September 2026

50 essays on geometry, discrete, topology, dynamics, computation, applied, probability, analysis, logic, number and algebra

23 August 2026

15 essays on geometry, discrete, topology, dynamics, probability, logic, computation, applied and algebra

22 August 2026

15 essays on number, geometry, probability, discrete, analysis and topology

Every essay, by subject · by subject · by thread · search