In Praise of Tetrahedra: Why Four Objects Change Everything

Four Is the First Hard Number

Plain-language explainer for doi:10.5281/zenodo.20581484 (#386)

The central idea in one sentence

Three objects can always be combined in a unique order; four objects cannot — and the amplitude for the ambiguity is the Wigner 6j symbol, whose geometry is a tetrahedron.

The puzzle

Take any physical objects — angular momenta, particles, representations of a symmetry group — and ask: does the order in which you combine them matter?

Two objects: combine $A$ and $B$. One way. No ambiguity.
Three objects: combine $A \otimes B \otimes C$. Two ways to bracket: $(A \otimes B) \otimes C$ or $A \otimes (B \otimes C)$. By the associativity of the tensor product, these are naturally equal. Still no ambiguity.
Four objects: combine $A \otimes B \otimes C \otimes D$. Three distinct bracketing schemes. These are isomorphic — they give the same answer eventually — but the isomorphism between them is non-trivial. It requires a new piece of data: the Wigner 6j symbol.

Four is the first number at which combining things in different orders genuinely produces different intermediate states. This is not a quirk of notation; it is a structural fact about any theory that combines more than three objects.

The tetrahedron carries the amplitude

The 6j symbol has six entries — one for each edge of a tetrahedron. This is not a coincidence.

The four objects being combined correspond to the four vertices of the tetrahedron. The three intermediate couplings (one for each bracketing scheme) correspond to three of the six edges. The four faces encode the triangle conditions that must be satisfied at each vertex. The symmetry group of the 6j symbol under permutation of its entries is $A_4$ — exactly the rotation group of the tetrahedron.

The Ponzano-Regge observation (1968) makes this precise: the 6j symbol is the amplitude associated with a tetrahedron, in the same sense that a Feynman diagram amplitude is associated with a graph. The tetrahedron is not a picture of the 6j symbol — it is the 6j symbol.

The same symbol in eight domains

The paper’s main claim is that the 6j symbol appears identically — not analogously — in eight physical domains:

Domain	What the 6j symbol does there
Nuclear spectroscopy	Pandya recoupling: particle↔hole conjugation
Quantum gravity (Ponzano-Regge)	Amplitude per chunk of spacetime
Topological phases (Levin-Wen)	Plaquette operator in string-net models
Conformal field theory	Crossing matrix for four-point functions
Knot invariants (Jones polynomial)	Amplitude at each crossing
3-qubit magic states	XOR-Fano win condition: $\lambda_L \in {+1,-1}$
Quantum networks	Entanglement swapping amplitude
3-body gravitational orbits	PSL(2,7) stability selection

In every case: same abstract operation (change the order of combining four things), same algebraic structure, different physical interpretation.

The Abelian case: trivial recoupling, non-trivial cohomology

For Abelian groups — in particular $(\mathbb{R}_{>0}, \times)$ and $(\mathbb{R}, +)$, the groups underlying financial mathematics — the 6j symbol is identically 1. Every irreducible representation of an Abelian group is one-dimensional, so all three bracketing schemes give the same result automatically. There is no recoupling ambiguity; the FLOP opcode evaluates to 1 and can be omitted.

This might suggest four objects are not special in the Abelian case. That conclusion is wrong.

The tetrahedron remains the critical geometric object — what changes is which feature carries the non-trivial content:

In the non-Abelian setting: the six edges carry the 6j amplitude (recoupling order matters).
In the Abelian setting: the four faces carry Čech residuals (pricing consistency matters).

For four financial institutions $A, B, C, D$, each of the four triangular faces carries a residual $c_{ABC}$, $c_{ABD}$, $c_{ACD}$, $c_{BCD}$ — how far that triangle’s pricing is from being globally consistent. The financial Pentagon identity asks whether these four residuals close:

\[c_{BCD} - c_{ACD} + c_{ABD} - c_{ABC} = 0\]

This is $\delta^2 \circ \delta^1 = 0$ — the coboundary of a coboundary vanishes. When all four faces are priced from a single model, this holds automatically. It can fail only when the faces come from independent sources. Its failure is a non-trivial $H^2$ class: systemic irresolvability that no bilateral instrument can fix.

Four is still the threshold. A triangle ($n=3$) cannot carry $H^2$. The hollow tetrahedron ($n=4$, topologically $S^2$) is the minimal complex that can. The non-trivial content has migrated from the edges to the faces, but the tetrahedron is still the right object.

The hierarchy in one table

$n$ objects	Physics (non-Abelian)	Finance (Abelian)
2	Trivial product	Bilateral rate / edge
3	Clebsch-Gordan (no ambiguity)	Triangle / $H^1$ residual
4	6j symbol (first non-trivial recoupling)	$H^2$ class (first non-trivial cohomology)
5	Pentagon equation (self-consistency of 6j)	Pentagon identity ($\delta^2 \circ \delta^1 = 0$)
$\geq 6$	Higher $nj$ symbols (software: sums of 6j products)	Higher simplices (software: built from tetrahedra)