Concept

Rank — where it appears

The number of independent directions a linear map's output can reach, which is the dimension of its image. It is the dimension of the image, and it plus the nullity is the dimension of the space the map started from.

Named by 15 essays across 5 fields — each of them below, with the objects they name alongside it.

A linear map redrawing the plane. The integer grid before and after a linear transformation; the shaded unit square becomes a parallelogram whose area is the determinant.

A matrix is a picture of what happens to the grid

Four numbers in a box is not an object anyone has intuitions about. The same four numbers, shown as an instruction for redrawing the plane, are.

algebra · Linear maps
A whole line arrives at the origin. A linear map whose determinant is zero, drawn before and after. One line of the plane is sent to the origin and the whole plane is sent onto another line; the dimension lost and the dimension kept add to two.

What a map throws away

A linear map redraws the grid, and the determinant measures how much it stretches area. When that measurement comes out zero the map has flattened the plane onto a line — and the question worth asking is not how much was lost but how much survived, because the two always add to what there was.

algebra · Linear maps
A wedge of 2 circles. Several circles all passing through one common point, each labelled with a generator, so that a loop is a word in those letters.

The subgroup that is freer than the group

A free group on two letters contains a subgroup of index three that is free on four. Nothing about a group makes that plausible; everything about a graph makes it obvious, and the argument is to stop looking at the group and start looking at the space whose loops it is.

topology · Covering spaces
What the map does to a circle. The unit circle with two perpendicular directions marked, and its image under [1.6, 1.2, −0.4, 1.1] — an ellipse whose axes are the images of those two directions, of lengths 2.04 and 1.10.

What a map does to a circle

Every linear map sends the unit circle to an ellipse. Two perpendicular directions go to two perpendicular directions, whatever the map is — even a map with no invariant direction at all, and even one that is not square.

algebra · Eigenvectors
Every fifth partition count divides, and the rank that says why. A row of partition counts with the ones in a congruence class marked, and a histogram of partitions sorted by rank.

Every fifth one divides

p(4) is 5, p(9) is 30, p(14) is 135, and every partition count at a number leaving four on division by five is divisible by five. Ramanujan read it off a table; the explanation is a way of splitting those partitions into five equal heaps.

number · Partitions
6 vertices folded to 4, and a graph that decides. The graph built from 3 generator words, folded until no vertex has two edges of one label leaving it. Reading a word from the base vertex decides membership, and 6 words are tested.

Folding a graph until it decides

A subgroup of a free group usually arrives as a list of words, and almost nothing about it is readable from the list. Draw the words as loops, merge every pair of edges with the same label leaving one point, and what is left is a machine that decides membership by reading.

topology · Covering spaces
2 independent rows and 2 independent columns. An array of 3 rows and 4 columns beside its transpose, with the independent rows of each shaded, showing the same count on both.

Counted across and counted down

A rectangular array has a number of independent rows and a number of independent columns. The two are counted in different spaces, from different objects, by computations that share nothing — and they are always the same number, which is why 'rank' is one word.

algebra · Linear maps
2 independent cycles and 3 independent cuts, on 5 edges. A small graph beside its incidence matrix, with the matrix's rank and nullity given and shown to be the number of independent cuts and the number of independent cycles.

The cycles and the cuts

The count that splits a map's source into what dies and what survives has nothing to do with graphs. Apply it to a matrix built from a graph's edges and points and it says that a graph's independent cycles and its independent cuts add to its number of edges — a theorem about drawings, obtained from an array.

algebra · Linear maps
A design on 7 points cannot have fewer than 7 blocks. The incidence matrix of a design on 7 points and 7 blocks beside the product of it with its own transpose, which has a constant off the diagonal and a determinant computed exactly.

More blocks than points

A schedule in which every pair meets once cannot use fewer groups than it has people. Nothing about the counting conditions says so, and the proof is not combinatorial at all — it is a determinant, computed over a field the schedules have nothing to do with.

computation · Finite geometry
How far a filter spreads a generator's output. Bars for bit depths 1 to 8: the ceiling ⌊24/v⌋ outlined, the raw generator's count of evenly spread consecutive outputs (24, 3, 3, 3, 3, 3, 3, 3), and the tempered count (24, 12, 6, 6, 3, 3, 3, 3).

A filter that changes only the spread

The generator most simulations use passes every output through a last scrambling step before anyone sees it. The step is reversible, it changes nothing about the period, and it cannot make the generator any less predictable. What it changes is which patterns of consecutive outputs can occur at all — on a small twisted generator, from half of them to every one.

computation · Pseudorandomness
Decoding from 20 + m random combinations: the chance it works. Bars for the exact probability that 20 plus m random binary combinations of 20 message bits have full rank, with simulated frequencies as dots, for m from zero upwards.

Erasures a code can see

If a channel loses bits instead of flipping them, and says which ones it lost, its limit rises from 1 − H(p) to 1 − p — and reaching it needs nothing cleverer than a random matrix and the solution of simultaneous equations. A random code needs, on average, 1.607 symbols more than the message it carries, whatever the message's length, and that number is a constant Erdős proved irrational.

computation · Error-correcting codes
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
A loop that is a hole, and the same loop filled in. a hollow triangle: 3 points, 3 edges, 0 triangles; holes in each dimension 1, 1, 0; a filled triangle: 3 points, 3 edges, 1 triangles; holes in each dimension 1, 0, 0.

A hole is a cycle that bounds nothing

A hollow triangle and a filled one have the same three edges and the same loop round them. In one the loop is the edge of something and in the other it is not, and that difference — a cycle that is not a boundary — is what a hole is. Counting holes is rank and nullity applied twice, and the Euler characteristic is what is left when the two applications cancel.

algebra · Linear maps
The expected rank a threshold rule achieves with the values shown, against what is known. n=1: 1.0000 (c=4.000); n=2: 1.2500 (c=1.000); n=3: 1.4009 (c=1.124); n=5: 1.5868 (c=1.257); n=10: 1.8141 (c=1.416); n=20: 1.9950 (c=1.555); n=50: 2.1557 (c=1.701); n=100: 2.2284 (c=1.781); n=200: 2.2725 (c=1.838); n=400: 2.2983 (c=1.877); n=800: 2.3130 (c=1.902).

The rank that remembers every value

Values arrive one at a time, each must be kept or discarded on the spot, and the aim is to keep one whose rank among all of them is low on average. Told only who is leading, the best rule gets 3.87. Shown the values, a rule gets below 2.33 — and how much lower the best possible rule goes is not known, because the rank of what is kept depends on every value seen, and the best rule may need to remember all of them.

probability · Optimal stopping
A Lights Out board and the presses that clear it. A 5 by 5 board with 11 lights on, cleared by 7 presses; it has four solutions in all.

The boards on which every light goes out

In Lights Out, pressing a light toggles it and its neighbours, and the goal is to turn every light off. Over the field with two elements the puzzle is a system of linear equations, and whether every pattern can be cleared depends only on whether one matrix is invertible. On a 3 × 3 board it is; on 5 × 5 a quarter of patterns can be cleared; on 39 × 39 one in four billion. Which sizes fail is decided by a common factor of two polynomials.

computation · Finite fields

Named alongside it

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

MatrixBasisKernelCounting argumentDeterminantDimensionFinite fieldGraphCovering spaceDecision procedureFree groupImage

All concepts