Answers are grounded in this paper and cite their sources.
Description: The four Wedderburn blocks of A_F = C ⊕ M2(C) ⊕ M3(C), of dimensions 1, 3, 8 and 24 — the second Jordan totients of the divisors of 6 — satisfy (Σd)² = 2Σd² − 4, that is 36² = 2·650 − 4. An earlier paper proved this midpoint identity unique to the seed (2, 3). This paper identifies it: it is Descartes’ circle theorem in the hyperbolic plane, whose constant is 4 where the Euclidean constant is 0, and the four blocks are four tangent circles. Measured in the half-plane below the line through the centres of the unit Ford circles, the Ford circle at p/q has geodesic curvature q² − 1 and the real axis is a horocycle of curvature 1. The block set is the configuration of the real axis and the Ford circles at 1/2, 1/3 and 2/5: the trivial block is the horocycle, the block n² − 1 is the Ford circle of denominator n, and the product block 24 is the circle at the mediant 2/5, of denominator 2 + 3. The product block equals the mediant’s block only at (2, 3), where the equality is Cassini’s identity F4 F6 = F5² − 1. For any seed (p, q) the midpoint defect is a Descartes quadratic in the product block: it is the product of the differences between the product block and the Ford blocks of the Farey mediant and the Farey parent. The mediant branch is the negative Pell equation u² − 2v² = −1 with u = pq + 1 and v = p + q, and the parent branch is the same equation for (p, −q). A squeeze between consecutive squares solves it, so over all integer pairs the identity holds only for {0, 1}, {1, 2} and {2, 3}: Ford’s own configuration, its first Farey step and the block set. One positivity certificate shows that among rank pairs the block set is the spacelike vector of least Lorentz norm, and that (2, 2) gives the only timelike one. Ford’s own configuration (the real axis, the line y = 1, F(0), F(1)) has curvatures (1, −1, 0, 0), and three Apollonian reflections reach the block set: 3 = 2·1 + 1, 8 = 2(1 + 3), 24 = 2(1 + 3 + 8). The second step is Pell’s equation 3² − 2·2² = 1. The Apollonian group of the Ford packing lies in the level-2 congruence subgroup of the extended Picard group. Pell’s automorph [[3, 4], [2, 3]] is its smallest pure boost, and the cube of the golden glide, [[1, 2], [2, 3]], the shortest translation found, with Doppler factor φ⁶; both carry Ford’s configuration onto a copy of the block set, as does every element of PGL(2, Z) with bottom row (2, 3). The golden unit 2 + 3φ = φ⁴ is the Doppler factor of a symmetry of the Ford packing that no Apollonian move reaches. The golden automorph and Pell’s automorph lie in the commutator subgroup of SL(2, Z), on which η² has trivial multiplier, and their traces (3, 3, 6) are Markov’s. The Apollonian moves that fix the real axis are exactly the congruence subgroup Γ(2); a boost of every allowed trace carries Ford’s configuration onto the block set, but among elements conjugate to Cohn matrices only Pell’s automorph does. The Jordan totients are Pisano periods: π(d) = J₂(d) for every d dividing 6, so the Fibonacci periods modulo 1, 2, 3, 6 sum to 36 = 6² (Jordan’s divisor sum) and their squares to 650, and π(n) = J₂(n) holds exactly when n divides 6. At a prime this says the golden ratio generates the multiplicative group of the field of p² elements, true only for p = 2 and 3. The top block 24 = π(6) = π(24) is the first nontrivial fixed point of the Pisano map; its factorisation is crossed — 3 from the prime 2, 8 from the prime 3 — unlike the 24 of the eta function, where 8 comes from 2 and 3 from 3. The ingredients — hyperbolic Descartes, Ford circles, the Apollonian group, the commutator subgroup of the modular group, Jordan totients, Pisano periods — are classical; the identification and the uniqueness results are new. Every statement is exact, nothing is fitted, and no physical constant is derived. Eighty checks in descartes_receipts.py. Lean 4 and Agda developments prove the identities, the classification over integer pairs and the least-norm statement, and verify the Pisano–Jordan theorem on finite ranges; verify_descartes650_formal.py restates every formal statement numerically (61 checks). Keywords: Descartes circle theorem, hyperbolic geometry, Ford circles, horocycle, Farey fractions, Apollonian circle packing, Apollonian group, Picard group, Lorentz group, Pell equation, golden ratio, Pisano period, Fibonacci numbers modulo n, Jordan totient, Dedekind eta function, Markov numbers, commutator subgroup, Wedderburn blocks, finite spectral triple, noncommutative geometry, negative Pell equation, Cassini identity, congruence subgroup, Lean 4, Agda, formal verification Copyright © 2026 Daland Montgomery. All Rights Reserved. Creative Commons Attribution Non Commercial No Derivatives 4.0 International For a private, proprietary license, contact: quidbit@icloud.com
No takes yet. Share an insight, caveat, or question.
Montgomery Daland (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: