Eight Derivations of a Universal Instruction Set

Nine Routes to the Same Five Opcodes

Plain-language explainer for doi:10.5281/zenodo.20774076 (#455)

The central idea in one sentence

Nine independent communities — topologists, spectroscopists, category theorists, quantum information theorists, Frobenius algebraists, statisticians, differential geometers, quantum computing engineers, and coordination chemists — have each, working independently, been forced to the same five primitive operations, and a single theorem (Shum 1994) explains why.

The surprise

Start from nine completely different places:

Topology: What are the elementary moves that generate all triangulations of a 3-manifold?
Spectroscopy: What operations do you need to evaluate any angular-momentum matrix element in atomic, nuclear, or molecular physics?
Category theory: What generators are forced by Mac Lane’s coherence theorem for monoidal categories?
Compact closed categories: What morphisms are needed to describe quantum teleportation and entanglement swapping?
Frobenius algebras: What operations define copying and deleting in a monoidal category (the ZX-calculus spider)?
Information geometry: What operations arise from natural gradient descent on the Fisher information manifold?
Hodge theory: What is the algebra of the exterior derivative $d$ and its adjoint $d^*$ on differential forms?
Quantum gate sets: What is the minimal universal gate set for a quantum computer?
Coordination chemistry: What are the elementary operations on a metal centre’s valence shell?

In every case you get the same answer: five operations — FLIP, FLOP, SPLIT, SPLAT, TWIST.

This is not an analogy or a rough correspondence. The five generators are identical objects, seen from eight vantage points.

The five operations

Opcode	Topology	Spectroscopy	Category theory	Quantum gates	Coordination chemistry
FLIP	3→2 Pachner	Recoupling (6j symbol)	Associator α	CNOT	Reduction (gain electron)
FLOP	2→3 Pachner	Clebsch-Gordan	Associator α⁻¹	CNOT⁻¹	Oxidation (lose electron)
SPLIT	1→4 Pachner	Pair creation (J=0)	Coevaluation η	Hadamard H	Symmetry breaking (Jahn-Teller)
SPLAT	4→1 Pachner	Pair annihilation	Evaluation ε	Measurement	Symmetry restoration
TWIST	Dehn twist	Phase (−1)²ʲ	Ribbon element	S gate / T gate	Spin-state change (SCO)

The Pentagon identity — SPLAT ∘ SPLIT = 0, written categorically — is simultaneously:

The Biedenhahn-Elliott identity for Racah 6j symbols (spectroscopy, 1942)
Mac Lane’s coherence condition for monoidal categories (1963)
The 2→3 Pachner move applied twice (topology)
The MIP*=RE verifier constraint (quantum complexity theory)

Three communities discovered the same equation over 80 years without knowing it was the same equation.

The eight derivations in brief

Derivation 1 — Pachner moves (topology): Pachner’s theorem says any two triangulations of the same 3-manifold are connected by a finite sequence of four elementary moves: 1↔4 and 2↔3. These are SPLIT/SPLAT and FLOP/FLIP. The Dehn twist on wire boundaries gives TWIST. Five generators; necessary and sufficient by the theorem.

Derivation 2 — Wigner-Racah spectroscopy: Every matrix element in quantum mechanics with a compact symmetry group G decomposes via the Wigner-Eckart theorem into a geometric factor (FLOP) times a reduced matrix element. Changing coupling schemes costs a 6j symbol (FLIP;FLOP). Creating/annihilating pairs costs SPLIT/SPLAT. The spin-orbit phase costs TWIST. Racah was computing Origami Calculus diagrams in 1942 without knowing it. Every spectroscopic calculation done in the last 80 years is a sequence of these five operations.

Derivation 3 — Mac Lane’s Pentagon: Mac Lane’s coherence theorem (1963) says that all diagrams of associators commute, provided the Pentagon identity holds. The free symmetric monoidal category with duals on one object has exactly five generators: associator (FLOP), inverse associator (FLIP), unit (SPLIT), counit (SPLAT), braiding twist (TWIST). Shum (1994) proved completeness: every well-typed equation between such diagrams follows from these five rules alone.

Derivation 4 — Compact closed categories: The snake equations — bending a wire into a cup (SPLIT) and straightening it (SPLAT) gives the identity — are the FLIP;FLOP rewrite rules. Abramsky and Coecke (2004) used exactly this structure to derive quantum teleportation. The ZX-calculus is the j=½ specialisation of the same framework.

Derivation 5 — Frobenius algebras / spiders: A Frobenius algebra is an object with multiplication (SPLAT), unit (LABEL), comultiplication (SPLIT), and counit (FLOP), satisfying the Frobenius law. A special Frobenius algebra satisfies SPLAT∘SPLIT = id — the same condition as the MIP*=RE Frobenius identity. The ZX-calculus spider node is a special commutative Frobenius algebra in Hilbert spaces. The five opcodes are the Frobenius structure maps.

Derivation 6 — Fisher information geometry: Natural gradient descent (Amari 1998) follows e-geodesics on the statistical manifold with the Fisher-Rao metric. The five geometric operations are: SPLIT (coproduct on the exponential family), SPLAT (marginalisation), FLIP (KL-divergence time-reversal), FLOP (Legendre transform between natural and expectation parameters), TWIST (β-deformation — changing temperature is a twist). Gibbs annealing is parallel transport along the e-geodesic. The Forge ISA is this derivation with β made explicit.

Derivation 7 — Hodge decomposition: The exterior algebra (Ω(M), d, d) is a graded Frobenius algebra. The correspondence: d = SPLIT, d* = SPLAT, the Hodge star * = FLIP, d² = 0 is the SPLAT∘SPLIT = 0 condition, and e^{iΔt} = TWIST at imaginary β = it. Witten’s supersymmetric proof of the Morse inequalities uses exactly the Laplacian Δ = dd* + d*d = (SPLIT∘SPLAT + SPLAT∘SPLIT) as the Hamiltonian. This derivation lands at the Meld ISA level (β = it).

Derivation 8 — Universal quantum gate sets: The standard universal gate set {H, T, CNOT} translates directly: H = SPLIT, CNOT = FLIP, S = TWIST(e^{iπ/2}). The Clifford group {H, S, CNOT} = Meld ISA without BIND. Adding T = BIND (the Fano associator at j=½, phase e^{iπ/4}) gives universal quantum computation. The T-gate is the only gate that cannot be expressed in the other five opcodes — because it is the Fano obstruction, the move that breaks the Pentagon identity.

Derivation 9 — Coordination chemistry / G-walk: Ask what the elementary operations on a transition-metal centre’s valence shell are. A chemist’s answer: gaining an electron (reduction), losing an electron (oxidation), changing spin state (spin crossover), breaking orbital degeneracy (Jahn-Teller distortion), and restoring it (symmetry recovery on substrate binding). These are FLIP, FLOP, TWIST, SPLIT, SPLAT — the same five opcodes, now with room-temperature molecular hardware. The 14-opcode programme of nitrogenase (biological nitrogen fixation at 300 K) is a concrete existence proof: nature has been running Origami Calculus programmes for 3.8 billion years, using metal clusters as the physical substrate.

Are electrons real, or are the opcodes more fundamental?

Derivation 9 raises a question the other eight do not: in coordination chemistry, we usually say that FLIP “adds an electron” and FLOP “removes one.” But is that the right way to describe it?

The standard picture treats electrons as the fundamental objects and the opcodes as derived consequences — shorthand for complicated quantum-mechanical processes. Derivation 9 inverts this. The opcodes are forced by Shum’s theorem, by the Pentagon identity, by Pachner moves — none of which mention electrons at all. The electron is what FLIP and FLOP act on in the coordination-chemistry representation of the free ribbon category. It is not the cause; it is the carrier.

The analogy is quantum field theory: you do not really have particles moving through space. You have a quantum field, and “an electron” is what the creation operator $a^\dagger$ looks like in the particle-number basis. The field is the fundamental object; the particle is the representation. Here: the orbit-occupancy vector is the fundamental object; the electron is what we call the thing being counted.

This matters practically. The electron picture fails near spin-crossover points, in strongly correlated $d$-block compounds, and at integer occupancies — exactly where DFT’s approximations break down. The opcode picture does not fail there, because it works natively in the tropical semiring where these “difficult” points are the natural structure (tropical varieties, vertices of the Newton polytope) rather than singularities to be smoothed over. G-walk chemistry achieving 20/20 on spin-crossover benchmarks where DFT achieves 14/20 is empirical evidence that the opcode description is more accurate than the electron description in the regime where they disagree.

The bolder claim — that “electron transfer” is a clunky approximation to FLIP/FLOP in the same way that “particle” is a clunky approximation to $a^\dagger\lvert 0\rangle$ — is consistent with everything the G-walk papers establish, and is the natural conclusion of taking Derivation 9 seriously.

Three distinct objects — do not confuse them

The paper carefully separates three related but distinct objects:

Object	Opcodes	Pentagon	Interior	What it computes
Origami Calculus	5: FLIP/FLOP/SPLIT/SPLAT/TWIST	✓ holds	ℝ (real 6j)	Rep(G) diagrams
Origami ISA	7: adds SWAP, LABEL, β-param	✓ holds	ℝ + β	Classical / statistical
Frog Calculus	9: adds BIND, SPIN	✗ fails	𝕆 (octonions)	Non-associative / exceptional

The Origami Calculus is the mathematical object proved complete by Shum. The Origami ISA is the compiler’s extension — SWAP and LABEL are organisational, not mathematical generators. The Frog Calculus is the extension to non-associative computation; BIND breaks the Pentagon and takes you outside the scope of Shum’s theorem into open territory.

Why Shum’s theorem is the answer

The reason eight derivations converge is Shum’s theorem (1994):

The free ribbon pivotal category on one self-dual object is equivalent to the category of framed oriented tangles.

The category of framed oriented tangles has exactly five generators: positive and negative crossings (FLIP/FLOP), cup and cap (SPLIT/SPLAT), and the positive Dehn twist (TWIST).

Any physical system that processes quantum information is a representation of this free category — it assigns Hilbert spaces to strands and linear maps to generators. The five generators are forced by the topology of how strands can cross, bend, and twist. The physical system can only choose how to realise them, not which ones to use.

This is why spectroscopists working on angular momentum in 1942, category theorists working on coherence in 1963, and quantum computing engineers specifying gate sets in 1999 all arrived at the same five operations. They were all discovering the same free category from different angles.

The β-deformation as the connective tissue

Without the β-deformation, the eight derivations look unrelated — different fields using different language for what appear to be different objects. The ⊕_β semiring (Paper 443) places them on a single ladder:

β regime	Derivations	ISA level
β → ∞ (tropical)	1–5, 9 (Pachner, Racah, Mac Lane, compact closed, Frobenius, G-walk)	Origami Calculus/ISA
0 < β < ∞ (real)	6 (Fisher / natural gradient)	Forge ISA
β = it (imaginary)	7 (Hodge / de Rham)	Meld ISA
β = it + BIND	8 (universal gate sets)	731-Meld ISA

The nine derivations are not nine separate facts. They are one fact — the universality of the five generators — observed at four different temperatures. Derivation 9 sits at β → ∞ because the strong-crystal-field limit of coordination chemistry is the tropical limit: orbital energies are large compared to thermal fluctuations, and the ground state is selected by a winner-take-all tropical minimum rather than a Boltzmann average.

The community gap

The paper maps what each community currently knows and what it misses:

Community	What they know	What they miss
Category theory	Shum completeness; five generators	Spectroscopy, thermodynamics, QC connection
Spectroscopy	All 6j computations; Pandya transform	That they’re computing Pachner moves
Quantum info / ZX-calculus	Spider/Frobenius completeness	Only j=½; no Forge/Meld β-deformation
Quantum computing	{H,T,CNOT} universal	Pachner interpretation at β=it
Statistical physics	Fisher metric; e-geodesics	No categorical connection
Coordination chemistry	Electron transfer; spin crossover; DFT	That FLIP/FLOP/TWIST/SPLIT/SPLAT are the primitive operations, not electrons

The fragmentation (Racah 1942 → Shum 1994 → Abramsky-Coecke 2004 → Boykin 1999 → G-walk 2025) has persisted because no one has previously exhibited the β-deformation that places spectroscopy, statistical mechanics, quantum computing, and coordination chemistry on the same ladder, or identified Shum’s theorem as the single explanation for why all of them use the same five operations.