The ABR Kernel

Same question, different names

The preprint’s correspondence table, drawn from the same data as the chips above. DFA = detrended fluctuation analysis · HRV = heart-rate variability · ML = machine learning. It calls the link “functional correspondence” rather than “mathematical equivalence”. (preprint, correspondence table)

How the preprint classifies each field

★ one of the six clearest cases, each, in the preprint’s words, “with qualifications noted for each”.

• Independent derivation: developed from first principles without demonstrable awareness of equivalent prior work.
• Qualified independent: the method is novel, but the researcher’s training or setting included significant exposure to the relevant physics.
• Domain transfer: an existing technique applied to a new domain, with traceable lineage.
• Intra-physics extension: new phenomena found with existing tools inside the same discipline (the definition says “intra-field”).
• Empirical precursor: critical-like behaviour observed before the formal framework existed.
• Foundational framework, foundational cross-domain: the general mathematics (scaling, self-similarity, power laws) that shaped several lineages; statistical physics “established the mathematics”.
• Prior cross-domain synthesis: an earlier work that already drew the fields together, a “textbook treatment”.

The preprint’s own limits

“The functional correspondence we document is phenomenological …” “Our citation analysis cannot rule out informal knowledge transfer. The quantitative evidence supports but does not prove independent derivation.” (preprint, Limitations)

Read the preprint, arXiv:2601.22389 · Go deeper: the cross-domain survey · Go deeper: bridges between domains

The formal table says locus or node for a point and edge or declared relation for an arrow. Provenance = the observation an arrow is traced to · projection = a declared stand-in (M or C) · admissible = traceable to an observable through M.

D := { x ∈ ℝⁿ | n < ∞, |x[i]| < ∞ }, the declared domain; M : O → D. “The kernel acts on M(o) only.” “No numerical result is itself an observable.” (kernel notation, line 7)

(kernel notation, lines 19–29; plain-language operators, lines 116–126, 164–181)

• Which kernel. “The minimal admissible interior is: s → t → u.” Without it, “the primary kernel Δ → Σ is the correct kernel”. The condition is structural, not a number reached by a ratio. (kernel notation, lines 53–57; plain-language operators, line 142)
• Before Δ or A. Admissible: uniform shift, declared unit scale. (kernel notation, line 33)
• Two phases. Phase 1 acts on one declared observation M(o). Phase 2 needs a declared pair: “Phase 2 requires the process, not the photograph.” (plain-language operators, lines 51–57, 231–237)
• Arrows. Provenance classes Continuation, Coupling, Persistence. Each arrow has exactly one admissible direction; the reverse needs its own observable provenance. (kernel reference, lines 330–337; kernel notation, line 13)
• C. A declared projection, not a kernel operator: state what it preserves and what it discards. The repository calls itself the ABR/ABRCE kernel (README, line 15). In the preprint’s supplement C is an operator and E = C(R(B(A(x)))) the composite; in the repository C is a declared projection, not a kernel operator. (kernel notation, lines 17, 31; supplement, lines 30, 63)
• Roles. Three project roles: Origin, the person who declares and holds responsibility; Generator, the one who proposes; Verifier, the one who checks, returning located findings, not approval, by asking: “Can I trace every mathematical operation continuously back to declared observables through declared operators?” The operators hold no agency. (role separation, lines 29–35, 150, 180; kernel reference, lines 131, 175)

Kernel V9, development branch

The Relational Primitive Provenance Invariant: “Every admitted value, operation, and result must have a continuous, verifiable provenance path to an Origin-declared observable relation within D.” It holds three things distinct: a provenance label, a declaration record and a calculation record. Four failure modes are named: missing, fabricated, disconnected and misattributed provenance. Milestones 2 and 3 (Milestone 3 declaring Req-1 to Req-7) were accepted by the project’s Verifier on 2026-10-24: internal milestones, not a release. The enforcement interface is OPEN, and the type names in the specification are working names. V8 constructors and operators are unchanged. (repository README, lines 37–66, 290–326; V9 specification, line 557)

Open conditions the repository lists

A-3, sums over an empty adjacency: OPEN. The ρ max-selection rule and its saturating form: OPEN, Origin declaration required. are_proportional edge-order dependence: HOLD. (kernel reference, lines 1031–1036)

The crate is Rust. cargo test runs 61 library tests and 3 provenance_demo tests on main; on kernel-v9 it runs 31 more conformance tests, with 4 ignored pending enforcement. (repository README, lines 81–87)

“The operators produce the same classes of output … at every declared region from Primary through Planetary. What changes across regions is the declared measurement mapping M, not the operators or their constraints.” (repository README, lines 20–23)

1. The preprint’s supplement records the formulation it describes under the name “Metatron Dynamics” (the framework name at the time; Metatron Dynamics, Inc. is the company that now holds the kernel’s copyright): four operators A, B, R and C on a ring with circular boundary conditions, composed as E = C(R(B(A(x)))). (supplement, lines 27–63)
2. The supplement says the framework “originated in distributed systems engineering”, that Macomber “independently derived” its operators, and that when Stephenson evaluated it in 2024 its behaviour at parameter boundaries “resembled critical phase transitions”. The preprint says that case prompted its survey. “Our framework is not the paper’s thesis; it is the accident that revealed the thesis.” (supplement, line 120; preprint, line 113)
3. The repository’s own history runs from persistence arrows (V5), through the primary kernel Δ → Σ with the ring ruled inadmissible (V6), the removal of rank and SVD quantities and of every kernel default (V7, V7.1), and one release number across every file (Kernel V8, 2026-09-26), to the declared arrow as the one starting point (Kernel V9 development branch, 2026-10-24). It starts at V5, the first row of the repository’s version table. (kernel reference, lines 1119–1136; repository README, lines 93–94, 334)
4. So the ring that the supplement’s formulation ran on is exactly what the current kernel rules inadmissible. The operators kept their letters, but their definitions and domain changed: A was a deviation from the mean and is now a directed difference, and C is no longer a kernel operator. (supplement; kernel notation, lines 19 and 31) This page states the difference and does not reconcile it.
5. Labels before Kernel V8 were “document-local … not kernel release versions”; this page uses them only as the repository’s history does. (repository README, lines 93–95; plain-language operators, line 280)

• It does not interpret or rank results: “The result is not an interpretation. It is a function of the declaration and the observable.” (plain-language operators, lines 239–241)
• It holds no value for ρ_base, χ₀ or cc (the coupling where two declared networks cross), and supplies no default. (kernel notation, lines 25, 29)
• Which kernel applies follows from the declared structure (s → t → u or not). (kernel notation, lines 53–57)
• It does not interpret what it computes; the chain ends at comparison with observables. (kernel reference, lines 101–102; plain-language operators, lines 239–241)
• A result is never an observable: “No numerical result is itself an observable.” (kernel reference, line 109)
• On the Kernel V9 development branch, the enforcement interface is not written yet. (repository README, lines 314–323)

And the preprint’s own limit: “The quantitative evidence supports but does not prove independent derivation.” (preprint, Limitations)