Cambrian Theorems
Recognition Physics Institute · the live run, and fifteen selected discoveries · live

Right now, a machine is reading the largest library of certified mathematics ever assembled, including the derivations of Recognition Science itself, and composing brand-new theorems that no human has written down. As of this reading it has certified 269,919. Every one was proved by the machine and certified by the Lean kernel, the same referee that checks human mathematics. Below the instrument are fifteen selected consequences from the run. Some are unconditional; others become true under premises stated beside them. Each formula is a theorem in the permanent record: not a suggestion, a proof.

269,919
theorems certified, and counting
33 / 148
of the library processed
0
free parameters in the theory behind them
campaign live
StageA: the frontier
New mathematics still growingyes
No padding with duplicatesholding
Failures nominalyes

Discovery one

The golden-ratio growth law

Sunflowers, pinecones, spiral galaxies: nature's favorite number is the golden ratio, 1.618… Recognition Science says the allowed scales of reality form ladders, and a ladder that keeps feeding back into itself must grow by one fixed factor from rung to rung. The machine proved that this factor is not arbitrary and not approximate: it is exactly the golden ratio, for every self-sustaining ladder, no matter how the ladder was built. The most famous constant in nature, derived from first principles, by a machine, on its own.

qS10_H2_1642 ∀ (S : GeometricScaleSequence) (k : ℕ), S.isClosed → scale S (k + 1) / scale S k = goldenRatio

grown from a harvested theorem × the theory's golden-ratio definition · shard 10 · kernel-certified

Discovery two

Every stable pattern's mass, no strings attached

The strongest form of the mass law the machine has found. Take any stable pattern of light in the theory and assume only one thing: its integer charge is not negative. Nothing else, no certificates, no tuning, no extra evidence. The machine proved the pattern's predicted mass is then forced into a single closed form: two raised to the pattern's sector power, times the golden ratio raised to its rung on the ladder, times the golden ratio plus that charge. Every stable pattern sits on the ladder, and the ladder needs nothing from us.

qS28_H1_1217 ∀ (ψ : LightPattern Λ), 0 ≤ ZOf ψ → predictedMass ψ = 2^(B_pow (sectorOf ψ)) · φ^(r0 (sectorOf ψ) + rungOf ψ − 14) · (φ + ZOf ψ)

the theory's stable-pattern structure × its φ-rung mass law, minimal hypotheses · shard 28, gen 1 · kernel-certified · jury significance 12/15

Discovery three

Fibonacci has no choice of ending

Take any list of positive numbers where each entry is the sum of the two before it, the rule behind 1, 1, 2, 3, 5, 8 and nearly every growth pattern in biology. If the list grows by a fixed ratio from step to step, the machine proved that ratio is forced into the theory's own constant, and it went one step further: the theory's recognition-cost curve, evaluated at its built-in energy scale, has its stiffness pinned to exactly that ratio plus one. The Fibonacci lock and the cost curve's stiffness turn out to be one fact, and the machine assembled it on its own.

qS06_H2_8743 ∀ (U : RSUnits), 0 < tau0 U → 0 < c U → ∀ (K : ℕ → ℝ), (∀ ℓ, 0 < K ℓ) → fibonacci_recurrence K → constant_ratio K (c U) → deriv (deriv (G ∘ costLambda (λkin U / τrec U))) 0 = c U + 1

the theory's Fibonacci lock × its cost functional equation · shard 6, gen 2 · kernel-certified · jury significance 11/15

Discovery four

Rest mass, computed from the bottom up

Feed the machine only the raw admissibility data at the bottom of a pattern's ladder, plus a nonnegative charge, and it assembles the pattern's rest mass from scratch. The answer lands in the same forced form: a power of two from the sector, a power of the golden ratio from the rung, and one factor of the golden ratio plus the charge. The machine found this route on its own, chaining a theorem it had discovered one generation earlier into the theory's mass law. Mass is derived here, not assumed.

qS22_H2_259 ∀ (ψ : LightPattern Λ) (E : BottomUpAdmissibilityEvidence ψ), 0 ≤ ZOf ψ → restMass ψ = 2^(B_pow (sectorOf ψ)) · φ^(r0 (sectorOf ψ) + rungOf ψ − 14) · (φ + ZOf ψ)

grown from a gen-1 harvested theorem × the theory's mass law · shard 22, gen 2 · kernel-certified · jury significance 11/15

Discovery five

A particle's mass, read off its own pattern

The theory's mass law predicts particle masses as golden-ratio powers with exact integer offsets, the formula behind the institute's mass papers. The machine proved the law is not a fit to data but a theorem about structure: take any pattern of light that is stable and closed, count the integer load it carries under the theory's averaging rule, and its rest mass is forced to be two raised to the pattern's sector power, times the golden ratio raised to its rung on the ladder, times the golden ratio plus that integer count. Mass is bookkeeping, and the machine showed the books balance.

qS06_H2_4946 ∀ (ψ : LightPattern Λ), StableClosedLightPattern ψ → SupportAveragedMassLawLoad ψ → 0 ≤ ZOf ψ → restMass ψ = 2^(B_pow (sectorOf ψ)) · φ^(r0 (sectorOf ψ) + rungOf ψ − 14) · (φ + ZOf ψ)

the theory's stable-pattern structure × its φ-rung mass law · shard 6, gen 2 · kernel-certified · jury significance 9/15

Discovery six

The cost of recognition has no dial

In most theories, how stiff the cost of doing something is counts as a free parameter, a number someone gets to tune. The machine proved the theory's recognition cost is different: run through the theory's built-in kinetic energy scale, the cost curve's second derivative is locked to exactly the square of the theory's own speed constant. There is nothing to adjust. The stiffness of recognition is the theory's number, not ours.

qS06_H1_1137 ∀ (U : RSUnits), 0 < tau0 U → deriv (deriv (G ∘ costLambda (λkin U / τrec U))) 0 = (c U)²

the theory's cost functional equation × its unit structure · shard 6, gen 1 · kernel-certified · jury significance 9/15

Discovery seven

A complete census cannot be faked

The theory derives exactly 2ᵈ patterns at each depth d. Suppose you walk through all of them in some order and ask whether the walk visited every parity class. The machine proved the answer collapses to one checkable fact: the route must be a perfect shuffle, never repeating and never skipping. Completeness of a census is exactly bijectivity of its enumeration, proved for every depth at once.

qS19_H2_3184 ∀ (d T : ℕ) (f : Fin T → Pattern d) (f₁ : Fin T → Fin T), Bijective f → Finite (Fin T) → (ParityClassCoverage (f ∘ f₁) ↔ Bijective f₁)

the theory's pattern coverage predicate × finite enumeration · shard 19, gen 2 · kernel-certified · jury significance 8/15

Discovery eight

A periodic table for meaning

Chemistry has a periodic table that says which atoms may exist. The machine certified the theory's analogue for signals: a stream of eight complex amplitudes, the theory's native signal shape, carries a definite forced meaning exactly when its mode belongs to one of the theory's token mode families, and the meaning it receives is the one the table assigns at the signal's golden-ratio level. Which meanings can exist is a table, not a coincidence.

qS29_H1_562 ∀ (v : Fin 8 → ℂ) (r : ℝ) (hr : 0 < r) (M : MeaningObject), Meaning v r hr = some M ↔ ∃ a : WTokenMode, forcedModeFamily v = some a ∧ M = tauModGauge (forcedSignature a) (forcedPhiLevel r hr)

the theory's meaning-forcing engine × its meaning periodic table · shard 29, gen 1 · kernel-certified · jury significance 8/15

Discovery nine

Real choice creates transcendental numbers

Some numbers, like pi, can never be pinned down by any polynomial with ordinary coefficients; mathematicians call them transcendental. The machine proved that the theory's own primitive, a specification that genuinely distinguishes two cases, is exactly the line between the algebraic and the transcendental: when an algebra's transcendence basis is indexed by such a specification, the algebra contains transcendental elements if and only if the underlying kind really can tell two things apart. Distinguishability, the theory's starting point, appears inside field theory as the boundary of algebra itself.

qS06_H2_8619 ∀ (K R A : Type) [CommRing R] [CommRing A] [Algebra R A], Nonempty K → ∀ (x : NontrivialSpecification K → A), Nontrivial R → IsTranscendenceBasis R x → (Nonempty {a : K // ∃ y, a ≠ y} ↔ Transcendental R A)

the theory's specifiability closure × field theory's transcendence bases · shard 6, gen 2 · kernel-certified · jury significance 8/15

Discovery ten

Full strength at the theory's own count

Take any grid of numbers you can undo, multiply it by its mirror image, and the result is always at full strength: nothing collapses, no information is lost. The machine proved this where it had never been proved before: for grids whose size is exactly 2ᵈ, the number of patterns Recognition Science derives at depth d. Two facts from two different worlds, linear algebra and recognition combinatorics, lock together in one statement, and the fit is exact.

qS03_H2_1866 ∀ (d : ℕ) (A : Matrix (Pattern d) (Pattern d) R) [ordered field R], IsUnit A → rank (A · Aᵀ) = 2^d

grown from a harvested rank theorem × the theory's pattern-counting theorem · shard 3 · kernel-certified · jury significance 7/15

Discovery eleven

The clock cannot move a settled mass

Take a settled matter model satisfying the zero-recognition-cost selection premise, then rotate its eight-beat anchor window by any number of ticks. The machine proved that the neutralized load does not move: after every rotation it remains exactly one eighth of the pattern's predicted mass. This joins the clock dynamics to the conditional mass-size law. The model class has an explicit inhabitant.

qS40_H1_619 ∀ (k : ℕ) (model : PhysicalSettledSigmaZeroModel3), normSq8 (neutralize ((cyclicShift^[k]) (pattern model).window₀)) = predictedMass (pattern model) / 8

clock-shift invariance × the settled sigma-zero mass-size law · shard 40, gen 1 · kernel-certified · zero-recognition-cost selection premise stated

Discovery twelve

Equivalent labels preserve the exact mass step

The mass ladder can label the same structural exponent in more than one way because rung and charge corrections trade against each other. The machine proved that any label equivalent to the one-rung successor inherits the exact ladder law: its predicted mass is the golden ratio times the prediction one rung below. The statement is conditional on the T6 mass-ladder bridge and the named rung-gap equivalence.

qS24_H1_1191 T6_To_CanonicalMassLadder_Bridge h6 → RungGapEquivalent rung₁ Z₁ (r + 1) Z → predict_mass s rung₁ Z₁ = φ · predict_mass s r Z

rung-gap uniqueness × exact φ-ladder scaling · shard 24, gen 1 · kernel-certified · bridge premises inhabited

Discovery thirteen

Reversal preserves the Z charge

Reverse the order of every accumulated phase in a recognition history. Under the existing Z-Theta Noether and death-readdressing certificates, the machine proved that the original Noether Z-charge equals the complexity of the reversed history. Re-addressing changes order without deleting accumulated recognition.

qS29_H1_951 ∀ (z : ZPattern), ZThetaNoetherCert → DeathReaddressingCert → Z_charge z = zComplexity ⟨z.accumulated_phases.reverse, 0⟩

Noether Z-charge identification × reversal conservation · shard 29, gen 1 · kernel-certified · both certificates inhabited

Discovery fourteen

Recognition time respects substitution and succession

The machine proved that two basic operations of arithmetic commute exactly with the theory's recognition-time interpretation. Substitute a term before interpretation and you get the same truth value as extending the environment by that term's value. Advance the successor formula at n and you get the original formula at n + 1. These statements are unconditional.

qS03_H1_1002 / qS03_H1_1003 msat recognitionTimeAlgebra ρ (subst 0 t φ) ↔ sat (Env.cons (eval ρ t) ρ) φ msat recognitionTimeAlgebra (Env.cons n ρ) (stepSucc φ) ↔ sat (Env.cons (n + 1) ρ) φ

recognition-time semantics × the arithmetic kernel's substitution and successor laws · shard 3, gen 1 · kernel-certified · unconditional

Discovery fifteen

The ledger cannot rewrite meaning-table addresses

Append any new time-offset entry to the ledger. The machine proved that every existing entry in the meaning periodic table remains retrievable at its original constructor index. New meaning can be recorded without changing the addresses of meanings already committed.

qS01_H1_276 ctorIdx x < enumList.length → (commit TauOffset.enumList e)[ctorIdx x]? = some x

append-only ledger addressability × the meaning table's constructor index · shard 1, gen 1 · kernel-certified

Why this matters

In ordinary mathematics a new theorem is a contribution to mathematics. Inside Recognition Science, a parameter-free candidate theory of everything, a new theorem is also a candidate fact about reality. These fifteen discoveries are about the theory's own machinery: the golden ratio forcing itself twice over, the mass law derived three independent ways, the cost curve's forced stiffness, a census that cannot be faked, the meaning periodic table, distinguishability itself surfacing as the boundary of algebra, and the pattern count locking into linear algebra. The machine is no longer just exploring mathematics. It is exploring its own theory of reality and coming back with receipts, and an independent jury now ranks what it brings home. The number at the top of this page is not a snapshot; it is a counter on a running campaign, and it updates while you read. The full instrument lives at the wave, and the judged admissions at Cambrian Wins.