The First Law

Distance kills entanglement. Exactly.


We ran the model. Twenty independent trees of depth 12, with 4,096 leaf particles each — 168 million particle pairs measured in total. The result is unambiguous.

Law 1 — The Decay Law
⟨MI(A,B)⟩ = C · exp(−α · dG)
Mutual information decays exponentially with genealogical distance.

The law holds across all 20 trees with a standard deviation of 0.004 in the exponent — three significant figures of universality. Twenty different random universes, all obeying the same decay law.

The measured exponent is α ≈ 0.890 at depth 12. The theoretical limit, derived analytically, is:

Exact Result
α = log(5/2) ≈ 0.916
Derived from first principles via Weingarten calculus. No free parameters.

Where log(5/2) comes from

The number log(5/2) does not come from fitting the data. It comes from solving an eigenvector equation for the average quantum channel over all possible Haar-random branchings. The branching tree has a transfer matrix, and α is the log of its dominant eigenvalue.

The calculation uses a branch of mathematics called Weingarten calculus — tools for averaging over random quantum operations — and produces an exact, closed-form answer. The key quantity is η2, the average Bell-sector amplitude of a single branching:

η2 = ⟨c(V)⟩ = 2/5   (exact)

The value 2/5 comes from the representation theory of the unitary group U(4). Each branching preserves 2/5 of the entanglement signal, on average. After dG/2 branchings on each side of a pair, the mutual information is proportional to (2/5)dG/2, giving α = log(5/2).

The 2.9% gap

The measured α ≈ 0.890 is 2.9% below the theoretical log(5/2) ≈ 0.916. This gap is fully accounted for by finite-depth effects — at depth 12, the tree has not yet reached its infinite-depth limit. The gap closes as depth increases, exactly as predicted. It is not a failure of the theory.

What makes this remarkable: α = log(5/2) is determined entirely by the dimensionality of the quantum system (2-dimensional qubits) and the structure of the Bell operator. It is a theorem, not a fit. There are no adjustable parameters.


Next: The Geometry →