Terms that vanish, a total that does not
harmonic is one function. Everything below came out of it during this
build, at parameters taken from the essays rather than invented for this page — so a figure
here is the same figure a reader meets in an essay, and if the generator changes, this page
changes with it.
With nothing chosen
Where a coin-signed geometric series lands, for λ = 1/3, 1/2, 1/√2, 1/φ
Each distribution is two shrunken copies of itself
More copies than there is room for, above one half
Two values of λ whose density can be written down
Four distribution functions, all continuous, one of them a staircase
What it checks while it draws
Collected by running the family and recording what it asserted, not written here. The count is how many separate times the claim was put to the test while these drawings were made.
- 1/1 + … + 1/2 is not a whole number ×190
- one number in 1..2 carries the most factors of 2 ×190
- the density is positive at 0.00 ×56
- run 3 of the q = 0.5 series is under its own bound ×24
- 1 golden signs land on F(4) − 1 points ×20
- the histogram bar at -2.47 matches the density ×19
- the histogram bar at 0.07 matches the density ×19
- the vector for λ^1 is λ^1 ×19
- the denominator of H(1) carries 2 to the power ⌊log₂ 1⌋ ×16
- H(2) is not a whole number ×15
- term 2 is under the envelope ×14
- the first 1 terms add to the closed form ×14
- φ's powers close in on whole numbers by a factor 1/φ at 3 ×14
- piece 1 is 0.50 of what was left ×13
- and p³ does not, for p = 5 ×9
- p² divides the numerator of H(4) ×9
- stage 0 has two to the 0 intervals ×8
- and the upper bound on block 0 is the k-th power of 2^(1−p) ×7
- and their lengths add to (2/3) to the 0 ×7
- block 0 sits between its own two bounds ×7
- block 1 reaches a half ×4
- H(4) lies strictly between 2 and 3 ×4
- and that bound is 1/(k log 2)^0.5 ×3
- between 50 and 2000 terms ×2
- the partial sum differs from 1/(1 − 2) by exactly 2 to the power of its number of terms ×2
- the tail of 1/n^1.1 past 100000 is bounded by an integral ×2
- the transform at λ = 0.65 dies away along its own powers ×2
- 2 is first passed at n = 4 ×1
- 3 at n = 11 ×1
- a ratio with even numerator has odd denominator ×1
- and between them they are all 32 strings of 5 bits ×1
- and by the last drawn stage it is already tiny ×1
- and by this stage they already cover most of the interval ×1
- and ends at one ×1
- and it closes on it ×1
- and it is still coming down towards the limit ×1
- and it is the largest power of 2 not above n ×1
- and lists none of them twice ×1
- and no address uses the digit one ×1
- and points of different classes are different points ×1
- and the denominator of the sum carries exactly that many ×1
- and the numerator is odd ×1
- and the remainder is the ratio to the power of the number of pieces ×1
- and they add to what the series says ×1
- and two of the series have ratios climbing to one, one of which converges and one of which does not ×1
- and what has gone is the geometric series it looks like ×1
- and what is left is exactly the tail ×1
- at 1/3 the count is the Cantor set's log 2 / log 3 ×1
- at one half every bar has height a quarter ×1
- at one half it is exactly one ×1
- at one half the distribution function is a straight line ×1
- at p = 0.4 it is large and growing with n ×1
- at p = 1 the variance left after the first tenth is tiny ×1
- at x = 0 every sign is plus and the value is H(digits) ×1
- at x = 1/3 the digits alternate and so do the signs ×1
- below one half most of the interval is empty (λ = 1/3) ×1
- Bertrand: a prime lies between 20/2 and 20 ×1
- between 12 and 40 rationals are covered ×1
- between 2,000 and 40,000 samples ×1
- between 3 and 20 random runs ×1
- between 3 and 7 ternary digits are listed ×1
- between 3 and 9 stages are drawn ×1
- between 8 and 16 binary digits ×1
- between eight and twenty-two terms are drawn ×1
- between four and twelve blocks are drawn ×1
- between ten and forty terms are drawn ×1
- between three and nine translates are drawn ×1
- between three and six exponents between 0.3 and 3 are compared ×1
- between three and six exponents between nought and three are compared ×1
- between three and ten stages are built ×1
- between two and five series the family knows are compared ×1
- between two and four values of λ ×1
- between two and six classes are drawn ×1
- by 4 it is more than ten times thinner than the normal curve ×1
- consecutive partial sums differ by the next term ×1
- each address names the left end of its own interval ×1
- each block leans further out than the one below it ×1
- each class is sampled at between 6 and 30 points ×1
- each rational is inside its own interval ×1
- every matrix drawn has its eigenvalues inside the unit circle ×1
- every partial sum drawn is under the head plus the geometric tail's total ×1
- every partial sum falls short of the limit ×1
- every partial sum sequence reaches (I − A)⁻¹ ×1
- every random run has settled down over its second half ×1
- every such sum has an even denominator ×1
- every triangle is isosceles with its two longest sides equal ×1
- every whole number from 2 up is crossed ×1
- exactly one number up to 12 carries the highest power of 2 ×1
- from one half up the whole interval is used (λ = 1/√2) ×1
- from one half up the whole interval is used (λ = 1/2) ×1
- from one half up the whole interval is used (λ = 1/φ) ×1
- inside the circle the partial sums settle on the inverse ×1
- it starts at nought ×1
- just below 1 every sign is minus ×1
- n runs to between 10 and 30 ×1
- no candidate length puts the union between one and three ×1
- no point is in two classes at once ×1
- no term is used twice ×1
- no two addresses read the same in binary ×1
- no two seeds is a small rational away from another, which would be a typo ×1
- on the circle they stay bounded without settling ×1
- one to four Pisot numbers this family knows ×1
- one to three of the roots 2, 3 and 4 ×1
- only one number up to 20 is a multiple of 19 ×1
- outside it they grow without bound ×1
- so 19 divides the denominator of H(20) exactly once ×1
- so its total is finite ×1
- so the translates of the selection do not overlap ×1
- the 2-adic digits times the denominator give the numerator, to 24 places ×1
- the alternating run is heading for log 2 ×1
- the answer is caught between an even and an odd partial sum ×1
- the bars carry all the probability ×1
- the bound on each block is exactly a half ×1
- the bounding series is geometric exactly when p is above one ×1
- the comparison holds a series from each side of the threshold ×1
- the comparison holds a series that converges and one that does not ×1
- the covered rationals leave no wide gap, so the covering meets every part of the line ×1
- the covering budget is between a hundredth and a half ×1
- the crossings of 2 to between 4 and 11 are drawn ×1
- the denominator of H(12) carries exactly 2^3 ×1
- the density at 0 misses 1/4 by a few millionths ×1
- the density at 2 is 1/8 to double precision ×1
- the density falls away from the centre ×1
- the diagonal matrix's powers only shrink ×1
- the distribution function climbs to one ×1
- the drawn lean is the sum of the drawn steps ×1
- the drawn running total is the partial sum ×1
- the entropy per sign has settled for the golden ratio ×1
- the entropy per sign has settled for the pentanacci ×1
- the entropy per sign has settled for the tetranacci ×1
- the entropy per sign has settled for the tribonacci ×1
- the enumerated bars match the closed form at λ = 1/√2 ×1
- the enumerated bars match the closed form at λ = 1/∛2 ×1
- the envelope dominates the series from some term onwards ×1
- the error is smaller than the last term used ×1
- the exponent is between a half and three ×1
- the first middle removed is between a third and an eighth ×1
- the function is constant across each removed interval ×1
- the function never decreases ×1
- the gap to ln n has settled on γ ×1
- the geometric series' ratio is the same at every term ×1
- the golden dimension is Alexander and Zagier's 0.99571 ×1
- the golden ratio's distribution has dimension below one ×1
- the golden transform holds level at 2πφⁿ ×1
- the halves share bars at λ = 0.6 ×1
- the halves share no bar at λ = 0.4 ×1
- the last term is small ×1
- the length left is the product of what each stage keeps ×1
- the lengths add to less than the budget ×1
- the list holds the rationals it was asked for ×1
- the longest surviving interval shrinks at every stage ×1
- the middle-thirds set of the same depth has far less left ×1
- the n-th root of the size of Aⁿ never falls below the spectral radius ×1
- the neglected tail is under an eighth of a bar ×1
- the neglected tail is under half of a bar ×1
- the other two grow a great deal before they shrink ×1
- the p = 1 row is the harmonic sum, which is a logarithm plus Euler's constant ×1
- the patterns are different ×1
- the patterns disagree only in a leading block ×1
- the pentanacci's distribution has dimension below one ×1
- the pieces and what is left fill the square exactly ×1
- the Pisot values sit just below the line ×1
- the plotted partial sums are the series ×1
- the prime is 2, 3 or 5 ×1
- the ratio is strictly between nothing and one ×1
- the rearranged total settles near the number it was aimed at ×1
- the removed intervals so far have the length the construction gives ×1
- the run is 1 to between 4 and 30 ×1
- the sample variance is near Σ 1/n² = π²/6 ×1
- the series is one the family knows ×1
- the signs are the ones the series has ×1
- the smallest interval is still wide enough to be represented ×1
- the special number is found by powers of two or by a prime ×1
- the staircase is flat over most of its interval ×1
- the sums run to between a thousand and a million terms ×1
- the sums run to between ten thousand and ten million terms ×1
- the surviving length is bounded away from nothing ×1
- the table runs to between 6 and 20 ×1
- the tail is drawn to between 3 and 4.5 ×1
- the target is visibly away from the sum the same terms give in order ×1
- the tetranacci's distribution has dimension below one ×1
- the total is not small ×1
- the triangle has side between 8 and 24 ×1
- the tribonacci's distribution has dimension below one ×1
- the two patterns land on the same point exactly ×1
- the union is no longer than the sum of the pieces ×1
- the view is one the family draws ×1
- there is one address for each surviving interval ×1
- this view is for a series that does converge ×1
- three to ten primes between 3 and 61 ×1
- two different translates never carry the same point of a class twice ×1
- two or three values of λ ×1
- two sign patterns of equal length ×1
- two to four exponents in (0, 1.5] ×1
- two to four values of λ ×1
- what is left and what has gone add to the whole interval ×1
- λ = 1/θ satisfies its relation for the golden ratio ×1
- λ = 1/θ satisfies its relation for the pentanacci ×1
- λ = 1/θ satisfies its relation for the tetranacci ×1
- λ = 1/θ satisfies its relation for the tribonacci ×1
- λ is between 0.2 and 0.9 ×1
Where it is called
Every figure on this list is drawn by the same rule, so a change to the rule changes all of them at once. That is why the list is published.
A coin in front of every power
Toss a coin for each sign of ±1 ± λ ± λ² ± … and the sum lands somewhere. Below λ = 1/2 it lands on a dust with gaps in it, at exactly 1/2 it lands anywhere with equal chance, and above 1/2 it lands on a smooth-looking hill — which for most λ has a density and for the golden value does not, though no picture can tell the two apart.
AnalysisA coin in front of every term
Put all plus signs in front of 1, 1/2, 1/3, … and the sum runs off to infinity; alternate them and it settles on log 2. Toss a fair coin for each sign instead, and the sum settles — every time, on a different number. Where it tends to settle has a smooth, flat-topped shape, and at the value 2 that shape takes a height that agrees with one eighth to forty-two decimal places and is not one eighth.
AnalysisA geometric series whose ratio is a matrix
1 + r + r² + … adds to 1/(1 − r) when r is smaller than one. Put a matrix in place of r and the same formula holds, with the inverse matrix in place of the fraction — but what must be smaller than one is not the matrix's size. It is its largest eigenvalue. A matrix whose eigenvalues are 0.9 and 0.8 can stretch vectors ten times over before its powers begin to shrink, and the series still converges, after a detour the eigenvalues say nothing about.
AnalysisA series that converges to minus one
1 + 2 + 4 + 8 + … runs off to infinity, and yet the formula for a geometric series says it should equal 1/(1 − 2) = −1. Measure size by how many factors of 2 a number has, instead of how large it is, and powers of 2 become small: the series converges, and to exactly −1. The same change of ruler explains why every repeating decimal is a fraction, and why repeating binary digits running off to the left are fractions too.
AnalysisA set that has no size at all
Slide the unit interval along itself by every rational and the points fall into classes. Choose one point from each and the resulting set has no length — not zero, not positive, none: countably many disjoint copies of it would have total length nought or infinity, and the union needs something in between.
AnalysisA staircase with no steps
A function that rises from nought to one, is continuous everywhere, and has derivative zero at almost every point. All of its climbing happens on a set of no length at all, which is possible because that set has uncountably many points.
AnalysisA sum whose terms vanish and whose total does not
Add a half, a third, a quarter, and keep going. The terms shrink to nothing and the total passes every number there is — but so slowly that no computation will ever watch it happen.
AnalysisAlmost none of it left, and still uncountably many
Remove the middle third of an interval, then the middle third of each piece left, and keep going. The lengths removed add to exactly the whole interval, so nothing measurable survives — and what survives can be paired off one for one with every point of the interval that was started with.
AnalysisCovering a set from outside
To say how long a set is, cover it with intervals and add their lengths, then take the smallest total any covering achieves. That definition is short, obviously right for an interval, and gives the rationals a length of nothing.
ProbabilityHow long until every one turns up
Draw at random from six equally likely kinds until all six have appeared. The wait is not six draws, and it is not sixty; it is fourteen point seven, and the number is a harmonic sum wearing a hat.
DynamicsInfinite on one side and nought on the other
Box counting returns a growth rate. Hausdorff's definition returns a measure — a quantity that is infinite for every exponent below the dimension and zero for every exponent above it, and the dimension is the one place where it is neither.
AnalysisNo interval in it, and length to spare
The middle-thirds set has no length because the removed pieces add to one. Remove shrinking middles instead and they add to a half — leaving a set that still contains no interval anywhere, and still has half the length it started with.
GeometryPinned between two sequences
The ring dissection makes the answer obvious and proves nothing. Archimedes' method proves it and makes nothing obvious — it never exhibits the area at all, it rules out every other value — and the recursion that drives it computes π by hand with one square root a step.
AnalysisThe same terms, in a different order, adding to whatever is asked
Flip alternate signs in the harmonic series and it converges. Reorder the terms — add nothing, remove nothing — and it converges to any number chosen in advance. Addition stops being commutative, and the picture shows where it goes.
AnalysisThe area that names the number
The number e can be defined without mentioning slopes at all. Slide right along the curve 1/x until the area underneath reaches exactly one, and stop. That is where e is, and the reason logarithms turn multiplication into addition is visible in the same picture.
AnalysisThe repair at the boundary
Where the geometric yardstick says nothing, compare a series with itself at doubled spacing. That one move turns every 1/n^p back into a geometric series, reads the threshold off at p = 1, and then produces an infinite hierarchy of boundaries with no slowest divergent series anywhere in it.
AnalysisThe series everything else is measured against
A geometric series is not one series among many. It is the yardstick: a total exists if its terms eventually fit under one, so a single comparison settles infinitely many questions — and the test built from it says nothing at all in exactly the place where the interesting cases are.
NumberThe sieve written as a product
Multiply out one geometric series for each prime and every whole number appears exactly once, as a single term. That identity turns a statement about factorisation into a statement about convergence, and it is where the analytic study of the primes begins.
AnalysisThe sum that fits in one square
Half, then a quarter, then an eighth, forever. Adding infinitely many things sounds like it should give infinity, and the picture that says otherwise is a square with a corner left uncut.
AnalysisThe sum that steps over every whole number
The harmonic sum 1 + 1/2 + 1/3 + … passes 2 at the fourth term, 3 at the eleventh, 4 at the thirty-first, and eventually every whole number there is. It never lands on one. The proof is a single number in the list 1, 2, …, n that carries more factors of two than any other — and the same arithmetic makes the numerators divisible by squares of primes they have no business knowing about.
AnalysisTwo sign patterns that land together
At λ = 1/φ the sign patterns + − − and − + + land in exactly the same place, because λ² + λ = 1. That one coincidence, repeated wherever it fits, puts 2ⁿ patterns onto a Fibonacci number of points, leaves the random sum's transform ringing at the same height forever, and makes a distribution that fills a whole interval live on a set of no length.
ProbabilityWhen to stop looking
Candidates arrive one at a time in a random order. Each must be accepted or rejected on the spot, with no going back and no way to know what is still to come. The best possible rule is to look at about a third of them and then take the first one that beats everything seen — and it works about a third of the time, however many there are.
AnalysisWhich functions can be added up
Riemann's integral works when the upper and lower sums close on each other. The exact condition for that, found once measure existed to state it in, is that the points where the function jumps have measure zero — which some nowhere dense sets fail.