The Fine-Structure Constant
α⁻¹ = 4π·11·exp(−w₈·ln(φ)/(4π·11)) is forced from J-cost uniqueness, φ self-similarity, and D=3; the certified band (137.030, 137.039) contains CODATA 137.036
Recognition Science derives the electromagnetic fine-structure constant α with no adjustable parameters. The closed form is α⁻¹ = (4π · 11) · exp(− (w₈ · ln φ) / (4π · 11)), where every term is forced by an upstream theorem: the cost functional J is the unique reciprocal-symmetric, normalized, calibrated solution of the d'Alembert / RCL equation (Law of Logic cost theorem, Aczél smoothness); φ is forced by self-similarity of the ledger; D = 3 forces the 8-tick cycle and the 11 passive edges of the cube; and the gap weight w₈ is the Parseval-normalized DFT-8 projection of the φ-pattern. The Lean verification certificate proves α⁻¹ ∈ (137.030, 137.039); CODATA 2022 measures 137.035999084(21), comfortably inside.
Predictions
| Quantity | Predicted | Units | Empirical | Source |
|---|---|---|---|---|
| α⁻¹ (RS closed form) | (4π·11) · exp(−(w₈·ln φ) / (4π·11)) |
dimensionless | 137.035999084(21) |
CODATA 2022 recommended value; arXiv:2502.09923 [physics.atom-ph]. |
| α⁻¹ band (Lean certified) | (137.030, 137.039) |
dimensionless | 137.035999084 |
Numerics.Interval.AlphaBounds (alphaInv_gt, alphaInv_lt). CODATA is inside the band. |
Derivation chain (Lean anchors)
Each row links to the corresponding Lean 4 declaration in the Recognition Science canon. A resolved anchor has a green check; an unresolved anchor flags a registry/canon mismatch.
-
1 J(x) = (x + 1/x)/2 − 1 is unique (T5 uniqueness) module unresolved
IndisputableMonolith.Cost.FunctionalEquation.law_of_logic_forces_jcost -
2 Closed self-similar ratio is φ theorem checked
IndisputableMonolith.Foundation.PhiForcingDerived.closed_ratio_is_phiOpen theorem → -
3 φ-forcing complete (T6) theorem checked
IndisputableMonolith.Foundation.PhiForcingDerived.phi_forcing_completeOpen theorem → -
4 8 = 2³ (T7) theorem checked
IndisputableMonolith.Foundation.DimensionForcing.eight_tick_is_2_cubedOpen theorem → -
5 8-tick cycle forces D = 3 (T8) theorem checked
IndisputableMonolith.Foundation.DimensionForcing.eight_tick_forces_D3Open theorem → -
6 f_gap definition (gap weight × ln φ) def checked
IndisputableMonolith.Constants.GapWeight.f_gapOpen theorem → -
7 w₈ from the 8-tick DFT structure def checked
IndisputableMonolith.Constants.GapWeight.w8_from_eight_tickOpen theorem → -
8 Geometric seed 4π·11 def checked
IndisputableMonolith.Constants.Alpha.alpha_seedOpen theorem → -
9 α⁻¹ = α_seed · exp(−f_gap/α_seed) (canonical resummation) def checked
IndisputableMonolith.Constants.Alpha.alphaInvOpen theorem → -
10 α = 1/α⁻¹ def checked
IndisputableMonolith.Constants.Alpha.alphaOpen theorem → -
11 α components witness lemma lemma checked
IndisputableMonolith.Constants.Alpha.alpha_components_derivedOpen theorem → -
12 Equivalent exponential form (44π factorization) theorem checked
IndisputableMonolith.Constants.AlphaExponentialForm.alphaInv_defOpen theorem → -
13 Theorem: 137.030 < α⁻¹ theorem checked
IndisputableMonolith.Numerics.Interval.AlphaBounds.alphaInv_gtOpen theorem → -
14 Theorem: α⁻¹ < 137.039 theorem checked
IndisputableMonolith.Numerics.Interval.AlphaBounds.alphaInv_ltOpen theorem → -
15 Structural provenance certificate def checked
IndisputableMonolith.Foundation.AlphaDerivationExplicit.alphaProvenanceCertOpen theorem →
Narrative
1. What needs to be derived
The fine-structure constant α governs the strength of the electromagnetic interaction in atomic and quantum-electrodynamic processes. Standard physics treats α as a free input with one of the most precise empirical values in all of science: CODATA 2022 reports α⁻¹ = 137.035999084(21). No mainstream framework derives that number from first principles. The RS claim is sharper: α⁻¹ is forced, no free parameter is available to tune, and the certified band (137.030, 137.039) is closed by Lean proofs that the CODATA value sits inside.
2. The closed form
The Recognition Science formula is
$$\alpha^{-1} ;=; (4\pi \cdot 11),\exp!\left(-,\frac{w_8 \cdot \ln \varphi}{4\pi \cdot 11}\right).$$
Equivalently, with $\alpha_{\text{seed}} = 4\pi \cdot 11 = 44\pi$ and $f_{\text{gap}} = w_8 \cdot \ln \varphi$,
$$\alpha^{-1} ;=; \alpha_{\text{seed}},\exp!\left(-,\frac{f_{\text{gap}}}{\alpha_{\text{seed}}}\right).$$
The Lean definition mirrors this exactly:
@[simp] def alpha_seed : ℝ := 4 * Real.pi * 11
@[simp] def alphaInv : ℝ := alpha_seed * Real.exp (-(f_gap / alpha_seed))
@[simp] def alpha : ℝ := 1 / alphaInv
with f_gap = w8_from_eight_tick * Real.log phi from
IndisputableMonolith.Constants.GapWeight.
3. Why every term is forced
3.1 The cost function J and the golden ratio φ
The Recognition primitives are reciprocal: pricing a forward step at $x$ and a backward step at $1/x$ must give the same cost. Imposing reciprocal symmetry, normalization $F(1) = 0$, the cost composition law (RCL), Aczél smoothness, and calibration $F''(1) = 1$ (in log coordinates) forces
$$F(x) ;=; J(x) ;=; \tfrac12(x + x^{-1}) - 1.$$
This is the Law of Logic cost theorem, formalized in Lean as
law_of_logic_forces_jcost in Cost.FunctionalEquation. It is the
load-bearing T5 link in the unified forcing chain, and was published
in the peer-reviewed Axioms (MDPI) 2026 as "Uniqueness of the
Canonical Reciprocal Cost."
Self-similarity of the ledger then forces the unique scale ratio to
satisfy $\varphi^2 = \varphi + 1$, i.e. $\varphi = (1 + \sqrt 5)/2$.
The Lean theorem closed_ratio_is_phi (and its companion
phi_forcing_complete) record this: any closed geometric scale
sequence satisfying the J-cost-derived composition law has ratio
equal to $\varphi$. There is no other choice consistent with the
primitives.
3.2 The 8-tick cycle and D = 3
The Recognition tick cycle has period $2^D$, where $D$ is the
spatial dimension. The fact that ledger states must come back into
recognition register after an integer number of ticks forces a
power-of-two period; the unique value matching the structural
constraints is 8, hence $2^D = 8$, hence $D = 3$. In Lean,
eight_tick_is_2_cubed proves $2^3 = 8$ by rfl, and
eight_tick_forces_D3 shows that any $D$ with $2^D = 8$ is exactly
$3$. The companion theorem power_of_2_forces_D3 chains the
arithmetic.
3.3 The integer 11
With $D = 3$, the canonical cell is $Q_3$, the 3-cube. Its boundary
$\partial Q_3 \cong S^2$ carries total Gaussian curvature $4\pi$ by
Gauss-Bonnet. Under the 8-tick projection, exactly one of the 12
cube edges is active per tick; the remaining 11 are passive. That
same integer 11 appears in $\Omega_\Lambda = 11/16$, in the
44-frequency baryon arithmetic ($44 = 4 \cdot 11$), and in the CKM
hierarchy. The cross-application of "11" is the discriminating
evidence that the count is structural and not freely chosen, and is
exactly the role of the alpha_seed = 4\pi \cdot 11 definition.
3.4 The gap weight $w_8$
The gap weight $w_8$ is the canonical Parseval-normalized projection of the DFT-8 of the φ-pattern onto a 64-cell ($8 \times 8$) grid weighted by $\sin^2(k\pi/8)$. Closed-form algebra plus Fibonacci identities for $\varphi^{-k}$ produces
$$w_8 ;=; \frac{348 + 210\sqrt 2 - (204 + 130\sqrt 2)\varphi}{7} ;\approx; 2.49057.$$
The integers $348, 210, 204, 130, 7$ are emergent from the algebra,
not chosen. The Lean module Constants.GapWeight records both the
projection definition and the lemma
w8_projection_equality that equates the projected and the
eight-tick forms.
4. The certified band
Numerical bounds on the closed-form expression are proved as Lean
theorems alphaInv_gt : 137.030 < alphaInv and
alphaInv_lt : alphaInv < 137.039 in Numerics.Interval.AlphaBounds.
Together they yield
$$\boxed{137.030 < \alpha^{-1} < 137.039}.$$
The CODATA 2022 value $\alpha^{-1} = 137.035999084(21)$ is inside,
with the experimental uncertainty four orders of magnitude tighter
than the RS band. This is a non-trivial empirical check: the
certified interval was set by the Lean proof, not by fitting to
CODATA. The structural provenance certificate
alphaProvenanceCert in Foundation.AlphaDerivationExplicit
records the chain in one place.
5. The equivalent exponential form
In Constants.AlphaExponentialForm, the theorem alphaInv_def
proves the algebraically equivalent presentation
$$\alpha^{-1} ;=; 44\pi ,\exp!\left(-\frac{8 \ln \varphi}{44\pi}\right),$$
matching the well-known "44π" notation. The two forms are the same number; the choice of $(4\pi \cdot 11)$ versus $(44\pi)$ surfaces whether the curvature term ($4\pi$ from $S^2$) or the recognition count ($44 = 4 \cdot 11$ frequency slots) is emphasised. The exponential structure follows from the canonical resummation of the recognition flow on the cube boundary.
6. What the framework predicts
A clean falsifiable consequence: any direct measurement of $\alpha^{-1}$ that lands outside (137.030, 137.039), at any frequency, in any energy regime, refutes RS. The CODATA 2022 value is well inside; the next factor-of-10 measurement-precision improvement on the muon $g-2$ or $h/m_e$ recoil experiments will narrow the empirical band further, but cannot move it outside the RS band without breaking the framework.
7. Where the derivation stops
The derivation stops at the structural primitives: ledger, tick, voxel, cost composition law, self-similarity. Below that level, the primitives themselves are forced by the unified-forcing-chain theorems T0 (Logic from Cost) through T4 (Recognition). T5 (J uniqueness) and T6 (φ forced) are the two load-bearing links above the primitives that feed into α; T7 (8-tick) and T8 (D = 3) close the geometry. Every step from T0 to the certified α band is a Lean theorem; the empirical seam lives at the experimental measurement of α⁻¹, not at any structural input.
Falsifier
RS would be falsified if either (a) the experimental value of α⁻¹ ever drifted outside (137.030, 137.039) by more than the experimental uncertainty, or (b) any of the upstream structural inputs (J uniqueness, φ forcing, D = 3, the 11 passive cube edges, the DFT-8 weight w₈) were shown to admit an alternative consistent with the Recognition primitives. The current measurement floor at 0.81 parts-per-billion (CODATA 2022) is more than four orders of magnitude tighter than the RS band, so this prediction is not loose: any future measurement-precision improvement that pushes α⁻¹ outside the certified band refutes the framework.
Depends on
References
-
paper
CODATA recommended values of the fundamental physical constants: 2022
doi:10.1103/RevModPhys.96.025002
Latest CODATA recommended value: α⁻¹ = 137.035999084(21). Used as the empirical benchmark. -
paper
Uniqueness of the Canonical Reciprocal Cost
Peer-reviewed source for the Law of Logic cost theorem that fixes J(x) = (x + 1/x)/2 − 1. -
standard
Recognition Science Lean library: IndisputableMonolith
https://github.com/jonwashburn/shape-of-logic
The full forcing chain, including J uniqueness, φ self-similarity, D = 3, and the α certificate, is checked in Lean 4 with zero sorry in the forcing chain modules.
How to cite this derivation
- Stable URL:
https://pith.science/derivations/fine-structure-constant - Version: 1
- Published: 2026-05-14
- Updated: 2026-05-14
- JSON:
https://pith.science/derivations/fine-structure-constant.json - YAML source:
registry/fine-structure-constant.yaml
Recognition Physics Institute. "The Fine-Structure Constant." Pith Derivations, v1. https://pith.science/derivations/fine-structure-constant.