Why does Econiac generalise beyond economics?

The mathematical core of Econiac — conservation laws, Gibbs dynamics, and H¹ cohomology — does not know it is doing economics. The same three structures appear in neuroscience, ecology, climate science, and metabolism. Econiac is the first fully-implemented domain; the others follow the same codebase.

The three pillars and where they appear

Econiac is built on three mathematical structures. None of them is specific to economics.

1. Conservation as ∂²=0

The Pacioli identity says: every financial claim has a matching counter-claim. Money can be created, but only symmetrically. This is ∂²=0 — the boundary of a boundary is zero.

The same identity holds, with different physical content, in:

Domain	What is conserved	The ∂²=0 statement
Economics	Financial claims	Every debit has a credit
Metabolism	Electron carriers (ATP, NADH)	Every oxidation has a reduction
Ecology	Energy through trophic levels	Every calorie eaten was a calorie somewhere
Climate	Atmospheric mass and energy	Every flux in has a flux out
Neuroscience	Charge at each neuron	Kirchhoff's current law at every synapse

In all five cases, ∂²=0 is not a model assumption — it is a constraint the system satisfies by construction. Models that violate it are wrong in the same way a balance sheet that doesn't balance is wrong.

2. Gibbs dynamics (β as decision sharpness)

The Gibbs lift replaces hard thresholds with smooth sigmoids parameterised by β — the inverse temperature. At β→∞ you recover the hard rule; at finite β you have a differentiable model calibratable from data.

This structure is not specific to financial agents. It is the universal model of a system that minimises free energy:

Karl Friston's free energy principle (the dominant framework in theoretical neuroscience) says that biological systems minimise variational free energy:

F = KL[q(s) ‖ p(s|o)] + log p(o)

where q(s) is the agent's internal model and p(s|o) is the generative model of the world. This is exactly the Gibbs distribution. The "precision" parameter in Friston's framework is exactly β in ours. High precision (high β) = sharp, cliff-edge responses. Low precision (low β) = gradual, exploratory responses.

The Gibbs lift is not borrowed from economics. It is the universal form of any system that balances exploration and exploitation under uncertainty.

3. H¹ cohomology as inconsistency

The sheaf H¹ signal measures whether local pieces of information fit together globally. It requires only: a network, local values at each node, and a consistency relation on each edge.

The same computation — same code, different calibration — applies across domains:

Domain	Network	Section	H¹ measures
Finance	Interbank exposure	Capital ratios	Bilateral solvency disagreement
Repo markets	Dealer-lender bipartite	Funding ratios	Roll-probability inconsistency
Neuroscience	Cortical regions	Prediction errors	Irreconcilable predictions
Ecology	Food web	Population ratios	Trophic inconsistency
Climate	Atmospheric cells	Temperature anomalies	Heat flux inconsistency
Metabolism	Reaction network	Metabolite concentrations	Stoichiometric imbalance
Photosynthesis	Chromophore graph	Energy efficiency	Broken Fano symmetry

The last row is already implemented: Paper 325 (FMO topological heat engine) uses the same sheaf_laplacian() function from econiac.finance.contagion to compute H¹ on the FMO chromophore network. The three-way isomorphism between the CHZ cascade, the repo run, and the FMO heat engine (Paper 334 §6) is empirical confirmation that H¹ is genuinely universal.

The Friston connection in detail

Karl Friston's free energy principle is the most ambitious unified theory in neuroscience. It claims that all biological self-organisation — perception, action, learning, evolution — can be understood as free energy minimisation.

In Econiac's language:

Friston	Econiac
Generative model p(s,o)	Restriction maps of the sheaf
Variational density q(s)	Section s of the sheaf
Free energy F = KL[q‖p]	H¹ signal = ‖L_F·s‖²/‖s‖²
Precision ω	Inverse temperature β
Active inference	Policy gradient ∂H¹/∂action
Markov blanket	The graph boundary ∂G
Surprise minimisation	Fixed-point iteration to H¹=0

This is not an analogy. The KL divergence and the sheaf H¹ signal are both measuring the same thing: how far a distribution (or a section) is from globally consistent with a generative model (or a sheaf). The mathematics is identical; the physical interpretation differs.

The practical consequence: Friston's framework lacks a concrete computational implementation that scales to large networks and admits policy gradients. Econiac provides exactly this. A thermology.neuroscience module that wraps the existing sheaf library with cortical region calibrations would be a direct implementation of the free energy principle at the network level.

Ecosystems

Trophic networks are conservation-law networks. Energy flows from producers through consumers to decomposers, with ∂²=0 at every node (energy in = energy out + stored). The Gibbs lift models predator switching behaviour (β measures how sharply predators switch prey species as relative abundance changes).

The H¹ signal on a food web measures trophic inconsistency — whether local population ratios are mutually reconcilable. This is a leading indicator of ecosystem collapse, for the same reason it is a leading indicator of financial cascades (Theorem 1, Paper 335): inconsistency accumulates before any individual population crashes through its threshold.

Keen's predator-prey model (econiac.economics.minsky) is already Lotka-Volterra dynamics — the same equations that govern debt-deflation spirals and species collapse. The calibration layer differs; the operator algebra is identical.

Climate

The atmospheric and oceanic circulation satisfies ∂²=0 (conservation of mass and energy). Tipping points — Amazon dieback, Atlantic overturning collapse, ice-albedo feedback — are cascade dynamics with Gibbs-like thresholds.

Paper 311 (Climate Yield Surface) already uses Econiac's framework for climate investment geometry. The extension to tipping-point early warning is direct: build a FinancialGraph where nodes are climate subsystems (Amazon, AMOC, Greenland ice, West Antarctic ice, permafrost), edges are their known physical couplings, and the section is the current anomaly. H¹ on this graph is a topological tipping-point indicator — detectable from existing observational data, model-agnostic, and (by Theorem 1 of Paper 335) a leading indicator.

Metabolism

Stoichiometric matrices in metabolic networks are ∂²=0 by construction. Flux Balance Analysis (FBA) — the dominant computational method in systems biology — already computes the null space of the stoichiometric matrix, which is H⁰ of the metabolic sheaf. H¹ of the metabolic network is the natural next step: it measures whether flux assignments are globally consistent, and its leading-indicator property (Theorem 1) predicts metabolic disease onset.

The Thermology unification

All five domains share the same three-layer architecture:

Layer 1: Conservation (∂²=0)
         — the network has a boundary operator
         — valid states satisfy the conservation law

Layer 2: Gibbs dynamics (β, free energy)
         — agents/nodes minimise free energy
         — β parameterises response sharpness
         — differentiable end-to-end via JAX

Layer 3: H¹ cohomology (sheaf Laplacian)
         — measures global inconsistency
         — leads cascade/failure by 2–3 periods
         — model-agnostic early-warning instrument

This three-layer architecture is Thermology — the mathematics of conservation-law networks with Gibbs dynamics. Econiac is the economics implementation. Thermion (three-body orbit search) is the physics implementation. A neuroscience implementation, ecology implementation, and climate implementation follow the same codebase with domain-specific calibration layers.

The shared library is thermology.core. Econiac and Thermion become domain-specific wrappers on top of it.