Every Computational Model Lives at One Point on the Same Map

Plain-language explainer for doi:10.5281/zenodo.21245459 (#543)

The central idea in one sentence

The inverse temperature β — normally treated as a positive real number controlling how “frozen” a system is — can be complex, negative, imaginary, or p-adic, and each choice places you at a different point on a single map called the adèlic β-plane, where every known computational model has a precise address.

A single equation with many faces

The Maslov-Gibbs Einsum is:

\[\pi_k(\beta) = \frac{e^{-\beta E_k}}{\sum_j e^{-\beta E_j}}\]

At $\beta > 0$ real this is the Boltzmann distribution — the standard tool of statistical mechanics. But $\beta$ does not have to be a positive real number. It can be zero, negative, imaginary, complex, or p-adic. Each choice gives a different computational model. All of them are the same equation.

The map

Draw the complex plane with Re(β) on the horizontal axis and Im(β) on the vertical axis. Every regime of computation occupies a specific region:

              Im(β)
                │
   Meld ISA ────┼──── quantum computation, unitary evolution
   (β = it)     │        PT-unbroken: gain/loss balanced
                │
  ──────────────┼──────────────────── Re(β)
  Inverted      │      Forge ISA → Origami ISA
  Origami       │      statistical mechanics → classical logic
  (β < 0)       │
  Lasers /      │      PT-broken: gain exceeds loss
  gain media    │
                │
  p-adic completions ℚ_p attached at the origin (totally disconnected)
Region β Arithmetic What it computes
Origin $0$ Uniform / smooth Hodge The Ambient — smooth containing manifold
Positive real $\beta > 0$ Gibbs weights Statistical mechanics; Forge ISA
$\beta \to +\infty$ Real, large Tropical $(\max,+)$ Classical logic; Origami ISA
Positive imaginary $\beta = it$ Complex amplitudes Quantum computation; Meld ISA
Negative real $\beta < 0$ Inverted Boltzmann Population inversion; lasers
$\beta \to -\infty$ Real, large negative Tropical $(\min,+)$ inverted Worst-case / max-energy path
Right half-plane $\beta = \sigma + it,\, \sigma > 0$ Damped amplitudes Dissipative quantum systems
Left half-plane $\beta = \sigma + it,\, \sigma < 0$ Amplified amplitudes Gain media; PT-broken QM
p-adic $\beta \in \mathbb{Q}_p$ Ultrametric / divisibility p-adic computation; Bruhat-Tits ISA

Nothing on this map is exotic mathematics. Each row is already a real physical or computational system. What is new is seeing them as one map rather than seven separate theories.

The positive real axis: from hot to frozen

Moving right along the positive real axis is familiar territory. At small β the distribution is nearly uniform — the system is “hot,” exploring all states equally. As β grows, the Boltzmann weight $e^{-\beta E_k}$ increasingly favours low-energy states. At $\beta \to \infty$ the distribution collapses to a single point — the ground state — and the arithmetic becomes the tropical $(\max, +)$ semiring. Boolean logic, classical computing, and the Origami ISA all live at this limit.

The snap event $\beta^\star$ — the threshold at which the system commits to its answer — is a phase transition on the positive real axis. Every classical algorithm, every optimisation, every catalytic cycle in chemistry is a journey along the positive real axis from the Ambient to the Origami ISA.

The imaginary axis: quantum computation

Rotating β by 90° — replacing β with $it$ — is a Wick rotation. The Boltzmann weight $e^{-\beta E_k}$ becomes $e^{-itE_k}$: a complex phase rather than a real weight. The sum over states becomes a sum of interfering amplitudes rather than a sum of probabilities. This is quantum mechanics.

The Meld ISA (β = it) is the quantum operative ISA. Classical, Clifford, and magic quantum computation are phase restrictions within the Meld:

  • Classical: phases discarded entirely (the $\beta \to \infty$ tropical limit of the Meld)
  • Clifford: phases restricted to fourth roots of unity ${1, i, -1, -i}$
  • Magic: algebraic irrational phases (T-gate: $e^{i\pi/4}$)
  • Full Meld: transcendental phases; complete Fano/octonion geometry

Quantum speedup — why Shor’s algorithm factors integers exponentially faster than any classical method — is the answer to: “what does turning on the imaginary axis of β add that the positive real axis lacks?” The answer is interference: paths with opposite phases cancel, allowing the algorithm to suppress wrong answers rather than merely making them improbable.

The negative real axis: population inversion

Going left along the real axis — negative β — the Boltzmann weight becomes $e^{+\lvert\beta\rvert E_k}$. Now higher-energy states are favoured. This is a population inversion: more of the system is in the excited state than the ground state.

This is not an abstraction. It is the operating principle of the laser. A gain medium (the lasing material) is pumped into a population inversion; stimulated emission then avalanches because the excited state is more probable than the ground state. Purcell and Pound demonstrated population inversion in nuclear spin systems in 1951 — the first experimental realisation of negative temperature.

In ISA language: negative β means the SNAP event crystallises to the maximum-energy orbit configuration rather than the minimum. The Origami ISA at $\beta \to -\infty$ selects the worst-case path — the tropical minimum in an inverted sense. This is the natural setting for adversarial problems: finding the hardest input, the worst-case schedule, the most unstable configuration.

The complex plane: PT-symmetric quantum mechanics

Between the real and imaginary axes lies the full complex plane. A value $\beta = \sigma + it$ with both parts nonzero gives amplitudes that both oscillate (from $it$) and grow or decay (from $\sigma$). This is a quantum system with gain and loss simultaneously.

This is exactly the setting of PT-symmetric quantum mechanics, pioneered by Carl Bender and Stefan Boettcher (1998). A PT-symmetric system has balanced gain and loss — the gain at one part of the system exactly compensates the loss at another. Such systems can have entirely real energy spectra despite having non-Hermitian Hamiltonians, as long as the gain-loss balance holds.

On the β-plane, PT-symmetric systems live in the band around the imaginary axis where $\lvert\sigma\rvert$ is small. There is a phase transition — the PT phase transition — at a critical value $\sigma^\star$:

  • PT-unbroken ($\lvert\sigma\rvert < \sigma^\star$): gain and loss balance; real energy spectrum; computation is stable
  • PT-broken ($\lvert\sigma\rvert > \sigma^\star$): gain overwhelms loss; eigenvalues come in complex conjugate pairs; the system amplifies without bound

At the PT phase transition itself, two eigenvalues coalesce — their eigenvectors become parallel — at an exceptional point. Exceptional points are branch points of the eigenvalue surface, and encircling one in parameter space applies a Berry phase that swaps the two coalescing modes. In ISA language: the exceptional point is the β-plane analogue of the snap threshold $\beta^\star$, and encircling it is a TWIST opcode — a topological phase correction.

The conjecture: the winding number around an exceptional point in the complex β-plane equals the TWIST eigenphase mod $2\pi$. If true, this gives a new topological invariant for PT-symmetric systems computable directly from the ISA programme.

The origin: the Ambient

The origin β = 0 is the Ambient — the smooth harmonic manifold that contains all the operative ISAs as limits. At β = 0, every state has equal weight; no commitment has been made; the system is maximally exploratory. The Ambient is described by Hodge theory: the operators $d$, $d^\star$, and the Laplacian $\Delta = dd^\star + d^\star d$ on the smooth manifold, finding harmonic representatives globally rather than making local decisions.

The Ambient casts all the shadows. The Origami ISA is the β → ∞ shadow; the Meld ISA is the β = it shadow; the Forge ISA is the shadow along the positive real axis. The Ambient is the object; the ISAs are its projections at different angles and distances.

The p-adic attachments: a different notion of closeness

Attached to the origin, but totally disconnected from the complex plane, are the p-adic completions $\mathbb{Q}_p$ — one for each prime $p$. These are not points on the complex plane; they are separate completions of the rational numbers that measure closeness by divisibility rather than by magnitude.

In $\mathbb{Q}_2$ (the 2-adics), 1024 = $2^{10}$ is very close to zero (highly divisible by 2), while 1/3 is far from zero. In $\mathbb{Q}_3$ (the 3-adics), 729 = $3^6$ is close to zero, while 1/2 is far. Each prime gives a different, equally valid notion of proximity.

The p-adic ISA runs the same Origami opcodes — FLIP, FLOP, SPLIT, SPLAT, TWIST, ORBIT — but over the $\mathbb{Z}_p$ arithmetic semiring rather than the real one. The computation is exact (no floating-point error), the “snap” is a discrete arithmetic condition rather than a continuous phase transition, and the geometry is the Bruhat-Tits tree — a $(p+1)$-regular branching tree — rather than a Euclidean simplicial complex.

By Ostrowski’s theorem, $\mathbb{R}$ and the $\mathbb{Q}_p$ are the only completions of the rational numbers. The adèlic β-plane — the complex plane together with all p-adic completions — is therefore the complete parameter space of computation. No other arithmetic is possible; Ostrowski’s theorem closes the map.

Why ℏ is not a fundamental constant

Planck’s constant ℏ is usually presented as a fundamental constant of nature — the scale that separates quantum from classical behaviour, whose value must be measured rather than derived. The β-plane perspective dissolves this mystery.

In the MGE, $\hbar = 1/\beta$ when $\beta$ is imaginary. The quantum-to-classical transition (ℏ → 0) is $\beta \to \infty$ along the real axis — the approach to the Origami ISA / tropical limit. Quantum mechanics (finite ℏ) is $\beta = it$ with finite $t$. Statistical mechanics is $\beta$ real and positive. All three are the same equation at different points on the β-plane.

ℏ is not a fundamental constant. It is a coordinate on the β-plane — the reciprocal of how far along the imaginary axis you are. Its numerical value in SI units reflects the choice of units (joules and seconds), not a deep fact about nature.

The Bender connection

Carl Bender, who pioneered PT-symmetric quantum mechanics, noted that the physical world seems to prefer PT-symmetric Hamiltonians even when they are technically non-Hermitian — because balanced gain and loss is physically natural (open quantum systems, optical cavities, waveguides with gain). The β-plane gives this observation a home: physical systems generically have $\beta$ slightly off the imaginary axis (some damping, some gain), and the question of which side of the PT phase transition they sit on is a question about the sign of Re(β).

The ISA programme of a PT-symmetric system is a Meld ISA programme with a small real-β perturbation. The PT phase transition is the boundary between computationally stable (PT-unbroken, real spectrum, reliable output) and computationally unstable (PT-broken, complex spectrum, runaway amplification) regimes.

The big picture

The β-plane is the unified parameter space of computation. Every computing model — classical Boolean logic, Gibbs sampling, quantum circuits, lasers, PT-symmetric waveguides, p-adic arithmetic — is the same MGE equation evaluated at a different address on this map. The differences between these models are not differences in kind but differences in location.

The Ambient (β = 0) is the smooth object at the centre. The operative ISAs (Origami, Forge, Meld, p-adic) are its faces at the boundary. The negative real axis (lasers, population inversion) and the complex plane (PT-symmetry, gain-loss systems) fill in the interior. Ostrowski’s theorem closes the map: there are no other completions of ℚ, so there are no other computational regimes.

The remarkable fact is not that these models are all related — it is that they are all the same equation, and the equation was already known. The MGE is not a new object. What is new is the map.

See also:

For the full technical treatment, see doi:10.5281/zenodo.21245459