Concept

Normal distribution

The bell-shaped distribution that sums of many small independent contributions settle into. It is the fixed point of convolution followed by rescaling, which is why it and no other shape is what sums arrive at.

Named by 21 essays across 7 fields — each of them below, with the objects they name alongside it.

A Galton board after 600 balls. 600 balls fall through 12 rows of pegs, each bouncing left or right at random, and pile up in a bell-shaped heap.

A bell curve assembled out of coin flips

Drop six hundred balls through a board of pegs, each bouncing left or right at random, and they pile up in a shape that can be predicted precisely. Nothing coordinated them.

probability · Central limit
Nine walks, and the square root. 9 independent walks of 400 steps, each step one place left or right. The dashed curves are ±√n: the walks stay near them, spill past them, and come back — which is what a typical distance means as opposed to a limit.

A walk that always comes home, until it does not

Step left or right at random, forever, and the walk returns to where it started with certainty. On a grid it also returns. In space it does not, and about a third of walks leave and never come back.

probability · Random walk
five values, unevenly weighted, and the mass outside 3 standard deviations. A distribution drawn as bars, with the windows one and a half, two and three standard deviations wide marked. The probability outside each window is summed and compared with the bound that knows only the variance.

How far from the average a thing can be

Knowing only an average and a spread — nothing about the shape, nothing about the number of outcomes, nothing about symmetry — the chance of landing three standard deviations out is at most one in nine. And there is a distribution that lands there exactly that often, so the bound cannot be improved.

probability · Concentration
One set of sums, two scalings, two different limits. The exact distribution of a sum of n independent copies, scaled two ways. Divided by n it collapses onto the mean; divided by the square root of n it holds a fixed width and settles into a shape.

The average settles and the wobble does not

Two theorems are usually met a page apart and sound as though one is a sharper version of the other. They are the same sums looked at through two different magnifying glasses: divide by the number of them and everything collapses to a point, divide by its square root and a shape appears.

probability · Central limit
One walk at three magnifications, and the shape it is heading for. The same random walk over three windows, each ten times longer than the last and scaled vertically by the square root of ten, so all three look alike. Beside them, the exact distribution of the position after a few step counts, standardised, closing on the bell curve.

The walk that becomes a curve

Shrink the steps of a random walk and it disappears. Shrink them while stretching the time in the right proportion — space by the square root of whatever time is divided by — and something is left behind, which is a curve nobody could draw.

probability · Random walk
A lopsided distribution added to itself, and the shape that returns. On the left, the exact distribution of a sum of copies of one lopsided distribution, standardised, for several counts: the shapes converge. On the right, the bell curve convolved with itself, which is the bell curve again.

The shape that averaging leaves alone

Adding independent quantities blurs their distributions together, and rescaling restores the width. Almost every shape is changed by that operation. Exactly one is returned unaltered, and that is why sums of unrelated things keep arriving at it.

probability · Central limit
Averages of a heavy-tailed quantity, which never settle. Running averages of draws from a Cauchy distribution, which jump rather than converge, beside the cumulative distributions of averages of 1, 4 and 16 draws, which lie on top of one another.

An average that never settles

The average of many independent quantities is supposed to steady as their number grows. For one famous distribution it does not steady at all — the average of a thousand draws has exactly the same distribution as a single draw, and no amount of further averaging changes it.

probability · Central limit
How fast a sum becomes a bell curve. The largest gap between the distribution of a standardised sum and the bell curve, against the number of terms, on logarithmic axes. Both summands fall along a line of slope about minus a half.

How fast the bell arrives

The limit theorem says a standardised sum approaches the bell curve and says nothing about when. The rate is one over the square root of the number of terms, the constant in front is made of the third moment, and both are visible.

probability · Central limit
A quadratic in the exponent, completed. Two panels sharing an x-axis. Above, the parabola −x² + 2x with its top at x = 1 marked. Below, e raised to that parabola: a bell centred at the same x = 1, with peak height e^1, beside the faint unmoved bell e^(−x²).

One number under every bell

The area under e^(−x²) has no formula in terms of the usual functions, and yet the area under e raised to any downward quadratic is known exactly. Completing the square in the exponent moves and squeezes every such curve into the same one, so a single number — √π — pays for all of them.

algebra · Completing the square
How often three candidates' majorities go in a circle. The exact probability that pairwise majority among three candidates is cyclic, for every odd number of voters from 1 to 41, under two models of how ballots are drawn, with their limits of about 8.77% and 6.25%.

How often the majority goes in a circle

Three voters and three candidates give 216 profiles, and 12 of them are cycles. Count every electorate up to 41 voters exactly and the share climbs towards 8.77%, a number Guilbaud found in 1952 as the solid angle where three half-spaces at the tetrahedral angle overlap. Add candidates and a winner goes missing half the time; let voters share one axis and cycles vanish. The number is always a property of the model of how ballots are drawn.

applied · Voting rules
How much of a sphere lies near its equator. Curves of the share of the sphere within ε of the equator against ε, for spheres in 3, 10, 100, 1000 dimensions: a straight line in three dimensions, a near step in a thousand.

A sphere that is nearly all equator

On an ordinary globe, the band within a tenth of the radius of the equator holds a tenth of the surface. On a sphere in a thousand dimensions the same band holds 99.85% of it, and the band of a fifth holds all but about two parts in ten billion. Almost every point of a high-dimensional sphere is near every equator at once — and so any function that cannot change quickly is, over almost all of the sphere, almost constant.

probability · Concentration
The polynomial bound falls exponentially behind the space. A log-scale plot against the dimension of the number of points of the space, the polynomial method's bound on a cap, and the known largest caps: the bound's line is less steep than the space's and pulls away from it.

The polynomial that bounds the caps

For forty years the best bound on a set of SET cards with no set among them shrank only like one over the dimension. In 2016 a two-page argument made it shrink exponentially, and the whole proof is a count of monomials: a table that is diagonal on a cap, one polynomial that describes it, and the fact that three parts of a degree cannot all be large.

computation · Finite fields
The volume of the unit ball and the area of its sphere, dimension by dimension. Unit ball volumes for dimensions 0 to 20, largest at 5 (5.2638); sphere areas largest at 7 (33.0734).

The ball that is largest in five dimensions

A disc of radius one has area π, a ball of radius one volume 4π/3, and in each further dimension the unit ball grows — until five, where its volume is 5.264, after which it shrinks towards nothing. The recursion that shows it is the ring dissection of the disc, done one dimension at a time, and the shrinking is not the ball getting small: it is almost all of a high-dimensional cube lying outside the ball, and almost all of the ball lying in a thin rind at its surface.

geometry · Circle area
The roots of random polynomials of degree 30 and 100: few are real. degree 30: 2 real roots at -2.335, -1.074; degree 100: 2 real roots at -0.327, 0.994.

Almost none of the roots are real

Pick the coefficients of a polynomial of degree a thousand at random, each one an independent draw from the bell curve, and ask how many of its thousand roots are real. The answer is about five. Mark Kac found in 1943 that the average grows only like (2/π) ln n, and the reason can be read off the real line itself: the real roots crowd towards +1 and −1 and spread evenly on a logarithmic scale of distance from them.

algebra · Polynomial roots
Five urns, each drawn as walks. Simulated walks of 2000 draws for urns with replacement matrices (0,1,1,0), (2,1,1,2), (3,1,1,3), (7,1,1,7), (1,0,0,1).

An urn forgets its start only below one half

Let each draw from an urn add balls of both colours in fixed amounts, and the long run depends on a single ratio of two eigenvalues. Below one half the urn behaves like a coin, its fluctuations spread like the square root of the draws and settle into a bell. Above one half the first few draws decide most of the outcome, the spread grows faster, and the shape that results is not a bell and depends on how the urn began.

probability · Random walk
Every slice of a tilted bell is a bell, centred off the axis. Contour ellipses of a two-variable bell with correlation 0.6, five vertical slices drawn as bells, their centres on the line y = 0.6x, and the ellipses' long axis on the diagonal y = x.

Regression is a square completed halfway

A bell in two variables, sliced at a fixed x, is a bell in y — and completing the square in y alone says where it is centred: at ρx, on a line flatter than the ellipse's own axis. That gap is regression to the mean, and the same completion says it runs backwards in time just as well as forwards.

algebra · Completing the square
The same bell on numbers, on primes less one, and on squares plus one. Standardised ω histograms: integers ≤ 2·10^7, p − 1 for p ≤ 2·10^7, n² + 1 for n ≤ 2·10^6.

The same bell on thinner sets

A typical whole number has about log log N prime factors, spread in a bell curve. That was a theorem about all numbers, built on the picture of each prime dividing independently. The numbers one less than a prime, and the numbers n² + 1, are far too thin for that picture to have any obvious claim on them, and on both the same bell appears — with means shifted by constants that come straight from how often each small prime divides them. The squares plus one that are prime, the case of a single factor, number 102,205 up to n = two million against 102,302 predicted; whether there are infinitely many is open.

number · Unique factorisation
A ring of identical clocks locked into a twist. Ring of 40, nearest-neighbour coupling, settled into a 2-twisted state; largest deviation 2.6e-11; order parameter 2.5e-16.

A ring of clocks settles into a twist

Put identical oscillators round a ring, each pulled only by its neighbours, and start them at random. They always lock — every one ends at the same frequency — but most of the time not in step: the angles wind round the ring a whole number of times and stay wound. The winding number a random start chooses is spread like a bell whose width grows like the square root of the ring, and only coupling to about a third of the ring makes full synchrony certain.

dynamics · Sensitive dependence
The correlations that reverse a coefficient. rho = 0.6: wedge shares 0.2051 and 0.2051 of the ellipse; arcsin(0.6)/pi = 0.2048.

Holding one still turns the other over

A quantity can rise with an outcome and still predict it downwards once a second quantity is held fixed. Completing the square in two predictors says exactly when: the pair of correlations has to fall in a wedge whose angle is arcsin of the predictors' own correlation. Averaged over every possible correlation matrix the chance is exactly 1/4 − 1/π², about one coefficient in seven; two coefficients never reverse together, and with two dozen predictors almost two in five do.

algebra · Completing the square
Shrinking beats the nearest point at every true mean. p=3: ratio at |θ|²/p = 0, 1, 5: 0.667, 0.863, 0.976; p=5: ratio at |θ|²/p = 0, 1, 5: 0.400, 0.732, 0.931; p=10: ratio at |θ|²/p = 0, 1, 5: 0.200, 0.621, 0.886; p=25: ratio at |θ|²/p = 0, 1, 5: 0.080, 0.549, 0.855

The nearest point is not the best guess

Measure ten unrelated quantities once each, with noise, and report the measurements. Nothing seems better. Yet pulling all ten towards nought by an amount the data choose gives a smaller total error whatever the true values are — in three dimensions or more, and never in one or two. The reason is a right angle: in many dimensions noise is nearly perpendicular to the truth, so a measurement is always too long, and the best correction is the shadow of the truth on it.

algebra · Inner product
How often each rule finds the Condorcet winner. n=3: 0.8824/0.9412/0.9559/0.6471; n=5: 0.8259/0.9751/0.9378/0.6866; n=7: 0.8281/0.9578/0.9278/0.6833; n=11: 0.8054/0.9656/0.9179/0.7079; n=21: 0.7941/0.9619/0.9097/0.7205; n=51: 0.7804/0.9624/0.9046/0.7339; limit 0.7572±0.0009/0.9629±0.0004/0.9011±0.0006/0.7563±0.0009; P(CW) limit 0.9116.

How often the rules find the majority favourite

When one candidate beats every rival head to head, a voting rule ought to elect it. Plurality does about three times in four when every ranking is equally likely, the runoff between the top two 96 times in a hundred, Borda's count 90. Every profile of up to 51 voters is counted exactly, the limit is sampled, and adding candidates pulls plurality down below one half.

applied · Voting rules

Named alongside it

The objects these essays reach for when they reach for this one.

VarianceScalingExpectationConvergenceIndependenceHeavy tailsLimitRandom walkSimulationCompleting the squareConvergence ratePi

All concepts