# Chapter ᚌ(G): The hall of mirrors *Teclam Áirme, the Mathematical Compendium — a copy for AI readers of https://www.teclamairme.com/#ch-T14, as of 2026-09-30. Links are relative to https://www.teclamairme.com/fortheaireaders/.* [Guide for AI readers](index.md) · [Its glossary cards](ch14-cards.md) · [Its references](ch14-refs.md) · [Previous chapter](ch13.md) · [Next chapter](ch15.md) *A footnote such as [r-14.3] is reference entry 14.3 and links to it. Section 14 of 28; chapters are named by their Ogham letter, never by number.* ᚌ – [Gort](ch14-cards.md#g-og-gort): associated with the ivy, Section 14 ## Chapter ᚌ(G): The hall of mirrors > “Symmetry, as wide or as narrow as you may define its meaning, is one idea by which man through the ages has tried to comprehend and create order, beauty, and perfection.” > — Hermann Weyl · *Symmetry*, Princeton University Press, 1952, p. 5 [r-14.22](ch14-refs.md#r-14.22) Formula: (([[x, [y, z]] + [y, [z, x]] + [z, [x, y]]](ch14-cards.md#g-jacobi)) + ([[x, x]](ch14-cards.md#g-liealt)))/(([dim 𝔰𝔲(2)](ch14-cards.md#g-su2) + [dim G₂](ch14-cards.md#g-g2) + [dim F₄](ch14-cards.md#g-f4) + [dim E₆](ch14-cards.md#g-e6) + [dim E₇](ch14-cards.md#g-e7) + [dim E₈](ch14-cards.md#g-e8))) {brace: [Lie theory](ch14.md#ch-T14-lie-theory)} · (([e₁ ∧ e₁](ch14-cards.md#g-wedge)) + ([e₁e₂ + e₂e₁](ch14-cards.md#g-cliffanti)))/([dim Cl(3)](ch14-cards.md#g-cl3dim)) {brace: [geometric algebra](ch14.md#ch-T14-geometric-algebra)} · (([1 + ζ₅ + ζ₅² + ζ₅³ + ζ₅⁴](ch14-cards.md#g-rootsum)) + ([Σ_g∈G χ(g), χ irreducible, χ ≠ 1](ch14-cards.md#g-charsum)))/(([|A₅|](ch14-cards.md#g-a5) + [|S₅|](ch14-cards.md#g-s5) + [26](ch14-cards.md#g-sporadic26) + [196883](ch14-cards.md#g-moonshine) + [|𝕄|](ch14-cards.md#g-monster))) {brace: [group theory](ch14.md#ch-T14-group-theory)} · (([disc((x − 1)²)](ch14-cards.md#g-disc)) + [[the general quintic is solvable in radicals]](ch14-cards.md#g-abelbracket))/([[ℚ(√2) : ℚ]](ch14-cards.md#g-degsqrt2) + [|Gal(ℚ(ζ₅)/ℚ)|](ch14-cards.md#g-galzeta5)) {brace: [Galois theory](ch14.md#ch-T14-galois-theory)} Key concepts: Lie theory, geometric algebra, group theory, and Galois theory: four views of symmetry. It includes symbols you may not have seen before, such as: [[x, y]](ch14-cards.md#g-liealg), [∧](ch14-cards.md#g-wedge), [ζ₅](ch14-cards.md#g-rootsum), and [𝕄](ch14-cards.md#g-monster). Zero here is symmetry canceling itself: roots of unity summing to nothing, a vector wedged with itself, and Abel’s proof that the general quintic has no formula in radicals. Try the mirror on this side. Left and right have swapped, yet the face looking back is still yours; this room is full of that idea. What can we change while keeping something recognizably the same? A shape may survive a reflection or a turn. An equation may survive a rearrangement of its roots. Symmetry lets those questions speak to one another. Rotate a regular pentagon through one fifth of a turn and its corners occupy the same set of positions. Rotate a sphere through a tiny angle and its shape is also preserved. Both are symmetries, but one belongs to a finite collection of allowed moves and the other to a continuous family. The mirrors have brought us in through reflection; the displays now give us other transformations to try. The four braces examine several ways of describing such structure. Field notes ### [Lie theory](ch14.md#ch-T14-lie-theory) {#ch-T14-lie-theory} Continuous symmetries can be studied through their [infinitesimal](ch26-cards.md#g-infinitesimal) motions. Start a small rotation about one axis, then another; reversing the order can change the result. In a matrix model the bracket [x, y] = xy − yx measures that failure to commute, and a *[Lie algebra](ch14-cards.md#g-liealg)* abstracts the rules such brackets obey. [r-14.1](ch14-refs.md#r-14.1) #### Numerator [[x, [y, z]] + [y, [z, x]] + [z, [x, y]]](ch14-cards.md#g-jacobi) = 0; the Jacobi identity, one of the defining laws of [a Lie algebra](ch14-cards.md#g-liealg). Three nested brackets, cycled through x, y, and z, always balance. With matrices you can watch it happen: multiply everything out and each of the twelve products meets its own negative. A Lie algebra keeps the rule even when there are no matrices to multiply. [[x, x]](ch14-cards.md#g-liealt) = 0; the other defining law: the bracket is alternating. Nothing fails to commute with itself: x then x is the same as x then x, so xx − xx = 0. #### Denominator Dimensions: the number of independent coordinates needed to specify an element of each algebra. [dim 𝔰𝔲(2)](ch14-cards.md#g-su2) = 3; the algebra of rotations, and of the two-state quantum systems in the next chapter. Three independent [infinitesimal](ch26-cards.md#g-infinitesimal) generators span this real [Lie algebra](ch14-cards.md#g-liealg). [r-14.3](ch14-refs.md#r-14.3) [dim G₂](ch14-cards.md#g-g2), [dim F₄](ch14-cards.md#g-f4), [dim E₆](ch14-cards.md#g-e6), [dim E₇](ch14-cards.md#g-e7), [dim E₈](ch14-cards.md#g-e8) = 14, 52, 78, 133, 248; the five exceptional simple Lie types, the famous outsiders of the classification. The finite-dimensional simple Lie algebras over the complex numbers, the ones that cannot be broken into smaller pieces, fall into four infinite families plus exactly these five exceptions. One of them returns later: [G₂](ch14-cards.md#g-g2) describes the symmetries of octonion multiplication, and comes back with the octonions in Chapter ᚖ(W). [r-14.2](ch14-refs.md#r-14.2) [r-14.20](ch14-refs.md#r-14.20) The first fraction is 0/528: 3 + 14 + 52 + 78 + 133 + 248 = 528. **Fine print.** This bracket is an algebraic operation, not an [Iverson truth bracket](ch08-cards.md#g-iverson). Both numerator terms are zero elements of the [Lie algebra](ch14-cards.md#g-liealg), read as the scalar 0, as always. The bracket must also be bilinear, respecting sums and scaling; the two laws in the numerator are the ones with a zero in them, not the whole definition. “Simple” means some bracket is nonzero and there is no nonzero proper ideal: no smaller piece, other than zero, that keeps every bracket taken with it inside itself. The sum needs nothing about the exceptional algebras but their dimensions; their cards say what each one is. ### [geometric algebra](ch14.md#ch-T14-geometric-algebra) {#ch-T14-geometric-algebra} Grassmann and Clifford let algebra carry oriented lines, areas, and volumes, so that products of directions become pieces of space. #### Numerator [e₁ ∧ e₁](ch14-cards.md#g-wedge) = 0; Grassmann’s exterior product, at its simplest. Let e₁ and e₂ be unit steps along two perpendicular directions. Their wedge e₁ ∧ e₂ stands for the parallelogram they span, an oriented piece of area; swap the order and the orientation flips, e₂ ∧ e₁ = −e₁ ∧ e₂. Wedge e₁ with itself and there is no parallelogram at all: one direction spans no area. [r-14.4](ch14-refs.md#r-14.4) [e₁e₂ + e₂e₁](ch14-cards.md#g-cliffanti) = 0; Clifford’s geometric product, which adds length and angle to Grassmann’s. For perpendicular unit vectors, e₁e₂ = −e₂e₁: swap them and the product changes sign; they anticommute. Their product is an oriented plane element, a *bivector*, which helps describe rotations in that plane. #### Denominator [dim Cl(3)](ch14-cards.md#g-cl3dim) = 8; the geometric algebra of ordinary space. Cl(3) has one scalar, three vectors, three bivectors, and one volume element: 1 + 3 + 3 + 1. Keep only the scalar and the three bivectors, and those four components multiply exactly like [the quaternions](ch22-cards.md#g-quaternions), the four-component arithmetic we meet properly in Chapter ᚕ(J). The algebra of oriented space already contains an arithmetic suited to rotations. [r-14.5](ch14-refs.md#r-14.5) The second fraction is 0/8. **Fine print.** The wedge and geometric products are related operations with different jobs. The anticommutation holds for orthonormal vectors in the chosen Euclidean Clifford algebra, Cl(3, 0), which the formula shortens to Cl(3). ### [group theory](ch14.md#ch-T14-group-theory) {#ch-T14-group-theory} A *group* records reversible moves and how they combine. A *representation* expresses those moves as matrices, and its *character* records each matrix’s trace. #### Numerator [1 + ζ₅ + ζ₅² + ζ₅³ + ζ₅⁴](ch14-cards.md#g-rootsum) = 0; the pentagon of the fifth roots of unity, a symmetry we can draw. With ζ₅ = e^2πi/5, the five powers sit at the corners of a regular pentagon centered at the origin, and their vector sum balances. Algebraically, multiplying the sum by ζ₅ − 1 gives ζ₅⁵ − 1 = 0; since ζ₅ ≠ 1, the sum must vanish. [r-14.6](ch14-refs.md#r-14.6) [Σ_g∈G χ(g), χ irreducible, χ ≠ 1](ch14-cards.md#g-charsum) = 0; a fundamental result of character theory, the pentagon’s balance on a much wider scale, taken here for G = A₅, the group that sits below the line. Take an irreducible representation of a finite group, one that cannot be split into smaller ones, other than the trivial one in which every move does nothing. Average all its matrices: the average can only land on vectors that every move leaves alone, and such a representation leaves no vector alone but zero. So the average is zero, and so is the sum of the traces. For a smaller group, the five turns of the pentagon, with each turn acting as multiplication by its own corner, the traces are the corners themselves, and the sum is the pentagon’s own 1 + ζ₅ + ζ₅² + ζ₅³ + ζ₅⁴. [r-14.7](ch14-refs.md#r-14.7) [r-14.6](ch14-refs.md#r-14.6) #### Denominator The counts move from shuffling five objects to [the Monster](ch14-cards.md#g-monster); 196883 counts coordinates in a representation rather than elements of a group. [|A₅|](ch14-cards.md#g-a5), [|S₅|](ch14-cards.md#g-s5) = 60, 120; the smallest simple group in which the order of moves matters, and all the shuffles of five objects. [S₅](ch14-cards.md#g-s5) holds all 5! = 120 ways to shuffle five objects; A₅ holds the 60 that take an even number of swaps. A₅ is also the group of rotations of an icosahedron, and an indivisible building block in the group-theoretic sense; it returns in the Galois brace. [r-14.8](ch14-refs.md#r-14.8) [26](ch14-cards.md#g-sporadic26) = 26; the sporadic simple groups, the exceptions to the infinite families. Every finite simple group belongs to one of several infinite families, except [twenty-six](ch14-cards.md#g-sporadic26). That sorting is the classification of finite simple groups, completed only in 2004. [r-14.9](ch14-refs.md#r-14.9) [|𝕄|](ch14-cards.md#g-monster) ≈ 8 × 10⁵³; the Monster, the largest sporadic group. The formula includes its exact order, a positive integer, which counts its elements. [r-14.10](ch14-refs.md#r-14.10) [196883](ch14-cards.md#g-moonshine) = 196883; chosen for [monstrous moonshine](ch14-cards.md#g-moonshineconj), one of mathematics’ great surprises. It is the size of the smallest nontrivial way to write the Monster’s moves as matrices: each of its elements becomes a grid of 196883 by 196883 complex numbers, so the number counts coordinates, not elements. The next integer, 196884, turned up somewhere else entirely, as a coefficient of the [j-function](ch14-cards.md#g-jinv), a modular function, and that neighboring pair helped launch [the connection between the Monster and modular functions](ch14-cards.md#g-moonshineconj). [r-14.11](ch14-refs.md#r-14.11) The third fraction is 0/([|𝕄|](ch14-cards.md#g-monster) + 197089): 60 + 120 + 26 + 196883 + [|𝕄|](ch14-cards.md#g-monster) = [|𝕄|](ch14-cards.md#g-monster) + 197089, a positive integer. **Fine print.** “Irreducible” means that no nonzero proper subspace is preserved by every transformation. The count of twenty-six follows the usual convention that files [the Tits group](ch14-cards.md#g-titsgroup) with the families rather than as a twenty-seventh sporadic. [r-14.23](ch14-refs.md#r-14.23) The Monster’s scale earns its place through the [moonshine connection](ch14-cards.md#g-moonshineconj), not simply because the integer is large. ### [Galois theory](ch14.md#ch-T14-galois-theory) {#ch-T14-galois-theory} Galois turned a question about formulas into a question about allowable rearrangements: the permutations of a polynomial’s roots that preserve their algebraic relations form its Galois group. Here the thing a mirror preserves is an algebraic relationship, with no reflected object to look at. #### Numerator [disc((x − 1)²)](ch14-cards.md#g-disc) = 0; the discriminant, the textbook detector of colliding roots. For x² − 2x + 1 = (x − 1)², the discriminant b² − 4ac gives (−2)² − 4 × 1 × 1 = 4 − 4 = 0: the root 1 is repeated. [r-14.12](ch14-refs.md#r-14.12) [[the general quintic is solvable in radicals]](ch14-cards.md#g-abelbracket) = 0; the Abel–Ruffini theorem, a famous impossibility, written as a false statement in an [Iverson bracket](ch08-cards.md#g-iverson). A radical formula uses arithmetic and root extraction. Quadratics, cubics, and quartics have one; the general quintic does not. Galois explains why. A formula exists exactly when the equation’s group can be taken apart, step by step, into pieces in which the order of moves does not matter; such a group is called solvable. The general quintic’s group is [S₅](ch14-cards.md#g-s5), and one of its pieces is A₅, the indivisible block from the group-theory brace, in which order does matter. So S₅ will not come apart that way, and no formula exists. [r-14.15](ch14-refs.md#r-14.15) [r-14.21](ch14-refs.md#r-14.21) [r-14.27](ch14-refs.md#r-14.27) #### Denominator [[ℚ(√2) : ℚ]](ch14-cards.md#g-degsqrt2) = 2; the simplest field extension. Add [√2](ch04-cards.md#g-sqrt2) to [the rationals](ch01-cards.md#g-rats) and keep everything arithmetic can build from them: every such number has the form a + b√2 with a and b rational, so two ingredients, 1 and [√2](ch04-cards.md#g-sqrt2), are enough, and the degree is 2. Besides doing nothing, this field has exactly one symmetry that leaves the rationals alone: it swaps √2 and −√2, a mirror in the algebra. [r-14.13](ch14-refs.md#r-14.13) [r-14.27](ch14-refs.md#r-14.27) [|Gal(ℚ(ζ₅)/ℚ)|](ch14-cards.md#g-galzeta5) = 4; the Galois group of the pentagon field from the group-theory brace. ℚ(ζ₅), the field built from ζ₅ and the rationals, has exactly four symmetries that leave [the rationals](ch01-cards.md#g-rats) alone: each sends ζ₅ to one of ζ₅, ζ₅², ζ₅³, or ζ₅⁴, shuffling the pentagon’s four other corners among themselves while the corner at 1 stays put. [r-14.14](ch14-refs.md#r-14.14) The fourth and final fraction is 0/6: 2 + 4 = 6. **Fine print.** Particular quintics can be solvable in radicals; the word *general* carries the claim. Galois’s criterion holds for fields like the rationals, of characteristic zero, which is the setting here. [r-14.27](ch14-refs.md#r-14.27) What this chapter adds to the formula: 0/528 · 0/8 · 0/([|𝕄|](ch14-cards.md#g-monster) + 197089) · 0/6 = 0 The hall has taken us from visible turns to transformations of things we cannot put in front of a mirror. One of them is about to reappear in a different key: the three-dimensional [algebra of rotations](ch14-cards.md#g-su2) we just counted is also the algebra of a two-state quantum system, where the geometry of directions becomes a language for possible outcomes. Can you hear that? It’s coming from the listening room. This journey not only has sights, it has sounds. Follow me! Watch and read - [Why this Equation Has No Formula (Galois Theory)](https://www.youtube.com/watch?v=VTGLyTPesIU) the clearest visual account of Galois’s idea: solving an equation means taking apart the symmetry group of its roots, and S₅ has a structure that obstructs any general formula in radicals - [Monster Group (John Conway) - Numberphile](https://www.youtube.com/watch?v=jsSeoGpiWsw) Conway himself on the largest sporadic group and the 196883 that started moonshine - [Galois theory](https://en.wikipedia.org/wiki/Galois_theory) Wikipedia - [Lie algebra](https://en.wikipedia.org/wiki/Lie_algebra) Wikipedia