Work

The Fine-Structure Constant from the Recognition Ledger

Year: 2026
Type: preprint
Abstract: With the matching between the ledger's lattice and the continuum derived at second order, under the record's statements and readings the paper names, and computed, the recognition ledger gives the fine-structure constant as one of two values, 1/α = 137.0244 ± 0.0072 or 137.0192 ± 0.0073, below the measured value by 1.6 and 2.3 errors. One statement the record does not make decides between them: whether a stretch's own term in the count of the links' own quanta, moved along that stretch's proper time, stops where the stretch reaches the length of its history's line, or follows the stretch into the line. Colour lands at −0.40 and −0.57 errors and the weak share at −0.17 and −0.63, and the three couplings' chi-square with their correlations rises from 1.40 to 4.79 and 12.06. If the 2026 lattice determination's perturbative piece is as our recomputation finds, the result is 137.0340 and 137.0288, 0.28 and 1.00 errors from the measured value; the paper keeps the determination's printed value. The fine-structure constant α ≈ 1/137 sets how strongly light and charge act on each other. It is measured to one part in ten billion; no accepted theory computes it. Here it is computed from the recognition ledger, with no continuous number fitted to light; the discrete readings it rests on were named with the measured couplings in view. At the ledger's cell, 2√(ln 2) Planck lengths on an edge, light's turn stiffness 1/e² = 1/(4πα) is a whole count, 8, times the stiffness a long wave reads from the ledger's own carrier, K = 1.040396 ± 0.000002. It is computed twice, independently, on the lattice of events with no measurement in it. Carried to rest by the Standard Model's computed running, 8K gives 1/α = 137.0244 ± 0.0072 or 137.0192 ± 0.0073 against the measured 137.035999177(21): 85 and 123 parts per million below, 1.60 and 2.32 of their errors. The matching's second order is derived from the record's statements with one working reading, that an exchanged link meets a history as the link that is read does, and, where the links' own quanta meet one another, a count the record does not state, besides the first order's readings; it is computed from published two-loop weights with no further choice. What the record does not fix in how the count of the links' own quanta treats a stretch at its line's length gives the two values; the tie in that count, and the top quark with the scalar, which is small, are in the error. Colour lands with the same dressing (−0.40 and −0.57 standard errors) but admits every read of the carrier; the weak share (−0.17 and −0.63) tests its own form, not the dressing, and two other forms land it too, one built from the carrier alone (−0.36 and −0.81); the quarks' factor of three (−0.63) rests on fitted terms. Light's read follows from what a coupling is, and the matching has no free constant if unrecorded histories are summed over every whole count: none at first order, and at second order the one the count gives. The counts follow from what an entry is, one comparison of its two sides priced once: other pricings miss, two events by thousands of errors, and the measured couplings over K, 8.0009 and 4.0019 under the first statement, 8.0013 and 4.0027 under the second, confirm it. It was first found with them in view. The other premises that carry the physics are readings identified with the couplings in view. The weight of the agreement is the coarse structure: light and colour at whole counts times one stiffness near 1.04 at a cell fixed by gravity, which a broad null gives once in 20 to 1,200. Counting every form tried for where the logarithms end, the precision adds little: some candidate fits all three couplings as well with chance 0.23 and 0.35; on the menus fixed before the count the chance is 0.026 and 0.055, falling to 0.00003 and 0.00008 as the derivations found afterward are granted. Three forms tried fit better under the first statement and four under the second, all of them walks on the bonds, which the record excludes. On the hadronic input Δα_had^(5)(M_Z²), which carries most of the error, the closure lands within 1.7 errors on four of the five lattice-based determinations under the first statement (one above its ask) and 2.4 below the fifth, and within 1.6 errors on three under the second (one on its ask) and 2.3 and 2.8 below the other two; it is strained or fails on every electron–positron evaluation with an error of 0.0001 or less, by 2.9 to 5.3 and 3.3 to 6.0 errors. The value it asks is filed as Δα_had^(5)(M_Z²) = 0.027882 ± 0.000019 (version 1.0 filed 0.027878 ± 0.000056); no determination made public since has scored it. With the second order computed it asks 0.027906 ± 0.000020 and 0.027944 ± 0.000020 under the two statements (version 1.8 filed 0.027939 ± 0.000048); on them the muon's leading hadronic part sits 2.3 and 2.8 errors high (a strain; a strain); on the perturbative piece as we recompute it, 1.3 to 1.4 and 1.8 to 1.9, inside the rule.
Language: English
Open Access status: green
OpenAlex ID: W7233957234
Cites: 56
Cited by:
Related to: 10