Pachner Moves and Quantum Symbols
The recoupling symbols of angular momentum theory (6j, 15j, …) are not merely algebraic bookkeeping devices. Each one is the transition amplitude of a specific Pachner move — the quantum weight of the higher-dimensional simplex that move sweeps out. This page makes that mapping explicit.
Background: What a Pachner Move Does
A Pachner move is a local replacement that changes the triangulation of an $n$-dimensional manifold without changing its topology. Every such move works by:
- Removing a set of $k$ simplices sharing a common face
- Replacing them with the complementary $\ell$ simplices that share the same boundary
The key insight is that the before and after configurations together form the complete boundary of an $(n+1)$-dimensional simplex. The quantum amplitude for that move is the state-sum weight of that $(n+1)$-simplex.
The 6j Symbol: Sweeping out a Tetrahedron
Context: 2D triangulations; 3D bulk gravity (Ponzano-Regge model).
The move (2-2 in 2D): Two triangles sharing an edge form a quadrilateral. The 2-2 Pachner move deletes the shared edge and inserts the opposite diagonal — one triangulation of the quadrilateral replaced by the other.
In the dual spin network: The two triangles are two trivalent vertices in an >--< shape. The 2-2 move changes the fusion channel from $s$-channel to $t$-channel. This is exactly the $F$-matrix recoupling operation of a fusion category — the associator $\alpha_{A,B,C}$.
The swept simplex: Place the before-quadrilateral and after-quadrilateral in 3D space and connect their corresponding vertices. The 3D volume enclosed between them is a tetrahedron. That tetrahedron has 6 edges, which carry the 6 angular momentum labels ${j_1, j_2, j_3, j_4, j_5, j_6}$ of the $6j$ symbol.
\[\begin{Bmatrix} j_1 & j_2 & j_3 \\ j_4 & j_5 & j_6 \end{Bmatrix} = \text{amplitude of the 2-2 Pachner move} = \text{weight of the tetrahedron}\]ISA placement: The $6j$ symbol is an H³ primitive — a 3-cocycle evaluated on a 3-simplex. The FUSE opcode (H²) is the associator that the 2-2 move enacts; the $6j$ symbol is the scalar amplitude of that move, which lives one cohomological degree higher. See opcodes.md for the FUSE / RECOUPLE distinction.
The 15j Symbol: Sweeping out a 4-Simplex
Context: 3D triangulations; 4D bulk gravity (Crane-Yetter / Ooguri models).
The move (2-3 in 3D): Two tetrahedra glued at a shared triangular face. The 2-3 Pachner move replaces them with three tetrahedra arranged around a single shared internal edge. (Two become three; the inverse 3-2 move is also valid.)
In the dual spin network: The network connecting the 5 tetrahedra (2 before, 3 after) requires 15 distinct angular momentum links to fully contract the tensor network of 4-valent intertwiners.
The swept simplex: The 4D hypervolume enclosed between the 2 initial and 3 final tetrahedra is a 4-simplex (pentachoron). A 4-simplex has 10 triangular 2-faces and 10 edges — its full amplitude is the $15j$ symbol.
\[\{15j\} = \text{amplitude of the 2-3 Pachner move} = \text{weight of the 4-simplex}\]ISA placement: H⁴ — outside the current Origami / Frog ISA. Appears in Crane-Yetter and Barrett-Crane spin-foam models of 4D quantum gravity.
The Master Identity: Biedenharn-Elliott
The recursion between dimensions is governed by the Biedenharn-Elliott identity, which states that the amplitude of a 3D Pachner move (which sweeps out a 4-simplex) factors into a sum over products of 6j amplitudes:
\[\sum_{x} d_x \begin{Bmatrix} \cdot & \cdot & \cdot \\ \cdot & \cdot & x \end{Bmatrix} \begin{Bmatrix} \cdot & \cdot & \cdot \\ \cdot & x & \cdot \end{Bmatrix} \begin{Bmatrix} \cdot & \cdot & \cdot \\ x & \cdot & \cdot \end{Bmatrix} = \begin{Bmatrix} \cdot & \cdot & \cdot \\ \cdot & \cdot & \cdot \end{Bmatrix} \begin{Bmatrix} \cdot & \cdot & \cdot \\ \cdot & \cdot & \cdot \end{Bmatrix}\](schematically: product of two $6j$ symbols = sum over product of three $6j$ symbols, matching the 2-3 Pachner move structure).
What this means for the ISA:
- Proving that a spin-foam model satisfies Biedenharn-Elliott is equivalent to proving it is invariant under 3D Pachner moves — i.e., that it is a well-defined topological invariant.
- In ISA terms: the $15j$ symbol (H⁴) decomposes into a circuit of $6j$ symbols (H³/RECOUPLE gates). This is why the $15j$ is not a new opcode but a depth-2 RECOUPLE circuit.
- By the same logic, the $9j$ symbol (which describes LS↔jj recoupling of four angular momenta) decomposes by Mac Lane coherence into a sum over products of three $6j$ symbols — it is a RECOUPLE circuit of depth 3, not an irreducible primitive.
The Complete Ladder
| Dimension | Pachner move | Swept simplex | Quantum symbol | ISA tier | Opcode |
|---|---|---|---|---|---|
| 1D (edges) | 1-1 (identity) | edge (1-simplex) | — | H¹ | TWIST |
| 2D (triangles) | 2-2 | tetrahedron (3-simplex) | $6j$ | H³ | RECOUPLE (proposed) |
| 3D (tetrahedra) | 2-3 | 4-simplex (pentachoron) | $15j$ | H⁴ | (Crane-Yetter; not proposed) |
Note: the Clebsch-Gordan / $3j$ symbol is the amplitude of a single triangle (2-simplex), not a Pachner move — it is an H² object (FUSE tier), the vertex amplitude before any move is executed.
Why 9j Is Not a Simplex
The $9j$ symbol has 9 angular momentum labels and describes the recoupling of four angular momenta (LS↔jj coupling in two-electron atoms). Despite involving four particles, it does not correspond to any simplex in the Pachner hierarchy:
- A tetrahedron has 6 edges → $6j$
- A 4-simplex has 10 edges → $15j$ (not $9j$; the mismatch is exact)
- No simplex has 9 edges
The $9j$ symbol is instead a graph amplitude — it corresponds to a specific Yutsis diagram (a graph with 9 edges connecting 6 nodes) that decomposes algebraically into a sum over products of three $6j$ symbols. It is a RECOUPLE circuit, not a RECOUPLE primitive.
Key References
- Ponzano and Regge (1968) — first identified the $6j$ symbol as the tetrahedral amplitude in 3D quantum gravity; the Ponzano-Regge model.
- Turaev and Viro (1992) — rigorous state-sum invariant of 3-manifolds using $6j$ symbols; the $6j$ as a 3-cocycle on a 3-simplex.
- Crane and Yetter (1993) — 4D state-sum model using $15j$ symbols; the $15j$ as the 4-simplex amplitude.
- Biedenharn and Louck (1981) — Angular Momentum in Quantum Physics; comprehensive treatment of $6j$, $9j$, and their identities.
- Kauffman and Lins (1994) — Temperley-Lieb Recoupling Theory and Invariants of 3-Manifolds; graphical calculus for $6j$ symbols.
- Barrett and Crane (1998) — relativistic spin-foam model; $15j$ amplitude as a Lorentzian 4-simplex weight.
- Paper 719 — orbital simplex: total torus-knot genus of atomic shell $n$ equals $\binom{n}{3}/2$; the $6j$ as the H³ primitive of the extended Origami ISA.
JLV diagrams, spin networks, and where Pachner moves actually apply
Three vocabularies describe the same objects, and conflating them causes real confusion. Worth stating explicitly.
JLV diagrams are spin networks
The Yutsis–Levinson–Vanagas graphical method (1960) and Penrose’s spin networks (1971) are the same mathematical object: a graph with SU(2) irrep labels on edges and $3jm$ intertwiners at trivalent nodes. Penrose arrived independently, from a different motive:
| year | what it was for | |
|---|---|---|
| JLV | 1960 | a calculational tool for atomic and nuclear spectroscopy |
| Penrose spin networks | 1971 | a foundational proposal about combinatorial spacetime |
Same diagrams, opposite intentions.
Correction (2026-08-01): an earlier version of this section said the two literatures “developed the identical formalism largely unaware of each other”. That is wrong. The connection is explicit and long documented: Yutsis diagrams embed into Penrose’s binor calculus, of which spin networks are the physical reading, and Penrose introduced his graphical calculus soon after the 1960 JLV book. Contemporary work treats JLV as foundational to the spin-network formalism, and SU(2) graphical calculus is used routinely for computing operator actions in the spin-network representation. The identification was made by the people who made it, not rediscovered here.
Spin networks do not need Pachner moves — spin foams do
This is the distinction that matters, and it is easy to get wrong:
| object | is | role | its moves |
|---|---|---|---|
| spin network | a labelled graph | a state (kinematics) | JLV rewrites: node sign change, arrow reversal, separation, cutting |
| spin foam | a 2-complex whose boundary is a spin network | a history (dynamics) | Pachner moves on the triangulation |
So: recoupling a spin network requires only the JLV moves. Pachner moves enter when one asks about evolution — when the triangulation itself may change.
And that is exactly why the two coincide
The bridge is the content of this page read backwards:
\[ext{Pachner 2--3} \;=\; ext{Biedenharn–Elliott} \;=\; ext{the pentagon relation on } 6j\]with Pachner 1–4 corresponding to the orthogonality relation. So a Pachner move on a triangulation, a $6j$ identity in angular-momentum theory, and the associativity coherence of a fusion category are one fact in three vocabularies — topological, spectroscopic, and categorical.
That is why JLV’s rewrite rules and the Pachner moves are not competing formalisms. JLV manipulates a network at fixed combinatorial structure; Pachner changes the structure; and the coherence conditions that make either well-defined are the same $6j$ identities.
And spin networks are already in ZX
More consequential than the history: the chain from spin networks to a categorical graphical calculus has already been published.
- East, van de Wetering, Chancellor and Grushin (2021), Spin-networks in the ZX-calculus, arXiv:2111.03114.
- Priestley (2025), Finite-Dimensional ZX-Calculus for Loop Quantum Gravity, arXiv:2511.15966.
ZX spiders are Frobenius algebras. So “SU(2) recoupling as a Frobenius structure” is not a new observation — it is the subject of an existing literature, and should be cited rather than claimed.
What that literature does not contain, and what therefore remains open to this corpus:
- a coefficient semiring k(β) — the ZX spin-network work carries no temperature;
- the occupancy boundary result (bond order is not recoverable from geometry);
- the basis-set-free two-centre overlap.
Those three are the honest contributions. The categorical framing of recoupling is not among them.
Consequence for the ISA
The Origami ISA’s Frobenius structure is not a synonym for JLV, despite the overlap. JLV is generated by a single self-dual trivalent node with no unit or counit; a Frobenius PROP has four generators and distinguishes multiplication from comultiplication. The honest relation is that SU(2) recoupling is an instance of the Frobenius structure, with the $3jm$ node as its fusion multiplication — so JLV inherits the general theory’s theorems rather than duplicating them.
References for this section
- Yutsis, Levinson and Vanagas (1962) — Mathematical Apparatus of the Theory of Angular Momentum; the graphical method.
- Penrose (1971) — “Angular Momentum: an Approach to Combinatorial Spacetime”; spin networks.
- El Baz and Castel (1972) — Graphical Methods of Spin Algebras.