> Архив сети: `01-research-reports/R2-nilpotent-ontology-review`

# R2 — Adversarial review: is (grading, nilpotent operator, metric) a novel minimal ontology?
Literature check, Aug 2026. Verdict: NOT novel; subsumed by Connes' spectral triple.

## The proposal reviewed
Primitives: (1) a grading indexed by integers, (2) one nilpotent d of degree +1
with d^2 = 0, (3) an inner product. Claim: d* is forced by the metric,
Delta = dd* + d*d is forced by both, harmonic = cohomology = "what stably exists",
Q = d + d* gives SUSY QM automatically, Witten index emerges. Sub-claim: an
operator and its adjoint are inseparable, so any duality of a symmetric and an
antisymmetric force is a theorem, not an axiom.

## 1. Closest existing framework: the spectral triple — and it does MORE
A spectral triple (A, H, D): involutive algebra A on a Hilbert space H, plus
self-adjoint D with compact resolvent and bounded [D, a]. Even triples add a
Z/2 grading gamma; real triples add an antiunitary J, signs fixed by KO-dimension
mod 8 (Connes, J. Math. Phys. 36, 6194, 1995).

Seven axioms: dimension, regularity, finiteness, reality, first-order condition,
orientability, Poincare duality in K-theory. See Connes, "Gravity coupled with
matter and the foundation of non-commutative geometry", Commun. Math. Phys. 182,
155 (1996), hep-th/9603053; and "On the spectral characterization of manifolds",
arXiv:0810.2088 (2008).

**Reconstruction theorem (2008):** a commutative unital spectral triple satisfying
the axioms with A = C^inf(X) determines and is determined by a compact oriented
Riemannian spin manifold — recovering manifold, metric, orientation, volume form
from spectral data alone.

**Connes distance formula:** d(p,q) = sup{ |a(p) - a(q)| : ||[D,a]|| <= 1 },
generalizing verbatim to noncommutative A.

**Spectral action** (Chamseddine & Connes, Commun. Math. Phys. 186, 731, 1997,
hep-th/9606001): bosonic action = Tr f(D/Lambda). Via heat-kernel expansion this
yields Einstein-Hilbert, a cosmological term, the Weyl term, and — for
almost-commutative geometry C^inf(M) x A_F — the full Standard Model
Yang-Mills-Higgs Lagrangian, with the Higgs as an inner fluctuation of D.

**Why our proposal loses the comparison:**
- it uses a DIRAC operator D (first-order, odd, encoding the metric via [D,a]),
  not a nilpotent d. d^2 = 0 is the wrong condition for encoding a metric.
- it needs FIVE primitives (A, H, D, J, gamma), not three. Our count of three
  undercounts: the metric does D's job and we omitted J and the grading is gamma.
- ours lacks the algebra A that lets Connes recover points, distances, gauge
  symmetry and matter.

**And the stronger version still hit a wall.** The spectral SM predicted a Higgs
mass ~170 GeV (big-desert assumption), excluded by the Tevatron in 2008. Measured:
m_H = 125.38 +/- 0.14 GeV (CMS, arXiv:2002.06398); ATLAS H->gamma gamma
125.22 +/- 0.11 +/- 0.09 GeV (arXiv:2308.07216). Chamseddine & Connes repaired it
in 2012 (JHEP 09 (2012) 104, arXiv:1208.1030) with a real scalar singlet sigma,
which "invalidates our previous prediction of Higgs mass in the range 160-180 GeV".
Lizzi (arXiv:1805.00411): adding such a boson by hand "lowers the predictive power
of the model"; with the extra parameter one can only state compatibility with data.

## 2. Hodge theory is a theorem — but the metric smuggles in nearly everything
Hodge decomposition (Riemannian): on a COMPACT oriented Riemannian (M,g),
Omega^k = H^k (+) d Omega^{k-1} (+) d* Omega^{k+1}, L2-orthogonal, with
H^k = ker Delta_k finite-dimensional and H^k ~ H^k_dR(M;R).
Rests on ellipticity of Delta and compactness (Rellich).

So yes, (grading, d, inner product) forces d*, Delta and harmonic = cohomology.
Standard, correct, elementary.

**But:** on a manifold the L2 inner product on forms IS the Riemannian metric plus
the Hodge star (needing an orientation) — the entire local geometry, curvature
included. Ellipticity and compactness are additional analytic hypotheses without
which Hodge theory is FALSE (it fails on non-compact manifolds). The metric also
fixes the dimension n via *: Omega^k -> Omega^{n-k}.
Calling it one third of the ontology dramatically understates it.

## 3. Nilpotent-operator-first physics is a 40-year programme
- **BRST**: nilpotent Q (Q^2 = 0), physical states = ker Q / im Q. Input: the
  entire classical gauge theory. Note (Figueroa-O'Farrill-Kimura): a nontrivial Q
  on a space with nondegenerate inner product FORCES an indefinite metric (ghosts).
- **Topological QFT of Witten type**: scalar 2Q with 2Q^2 = 0, L = {2Q, V} making
  correlators metric-independent (Witten, Commun. Math. Phys. 117, 353, 1988).
  Input: the twisted N=2 SYM action, gauge group, manifold.
- **AKSZ**: BV action from a nilpotent homological vector field Q on a graded
  target. Input: the target Q-manifold and symplectic form.
- **Batalin-Vilkovisky**: nilpotent s on field-antifield space; master equation
  (S,S) = 0. Input: fields, antifields, classical action.
- **String field theory** (Witten, Nucl. Phys. B268, 253, 1986):
  S = (1/2) Int Psi * Q Psi + (1/3) Int Psi * Psi * Psi — a differential graded
  associative algebra, exact analogue of (forms, d, wedge, Int). EOM
  Q Psi + Psi * Psi = 0 is a Maurer-Cartan / zero-curvature equation. For the
  superstring the star product is non-associative and a genuine A-infinity
  structure is needed (Erler, Konopka, Sachs, JHEP 04 (2014) 150, arXiv:1312.2948).
  Input: the worldsheet CFT that DEFINES Q and *.

**Pattern:** the nilpotent operator organizes physics; it never produces the
action, matter content, or numbers.

## 4. Witten SUSY QM — see R1 §7. Worldline SUSY, generic, says nothing about
spacetime superpartners.

## 5. The curvature obstruction — THE decisive objection
A connection gives a complex only if flat: d_nabla^2 = F, so d^2 = 0 iff F = 0.
With curvature the correct objects are:
- **Curved dg-algebras** (Positselski 1993): degree-+1 derivation d plus a
  curvature element c of degree 2 with d^2(x) = [c,x]. Keller, Lowen & Nicolas
  (J. Pure Appl. Algebra 214, 1271, 2010, arXiv:0905.3845) state it verbatim:
  such objects "have no cohomology, and there is no classical derived category";
  over a field the only derived category is ZERO (their Prop. 3.7).
- **Curved A-infinity algebras** (Getzler-Jones; Nicolas, arXiv:math/0702449):
  m_0 != 0 obstructs m_1^2 = 0; "it is not clear how to define the cohomology".
- **Matrix factorizations** (Eisenbud 1980; Kontsevich; Orlov): here one EMBRACES
  Q^2 = W*1. Physically the B-type D-branes of Landau-Ginzburg models
  (Kapustin-Li). The invariant is the singularity category, not cohomology.
- **Twisted complexes** (Bondal-Kapranov).

**So "harmonic = what exists" is valid only in the flat/free sector.** Turn on
interaction and the cohomological ontology must be replaced or evaporates.
Nilpotency is not stable under the deformation physics requires.

## 6. Discrete version — the same failure mode
**Eckmann (1944):** on a finite simplicial complex, L_k = d_{k+1} d_{k+1}* +
d_k* d_k gives ker L_k ~ H^k(K;R) and a discrete Hodge decomposition.
k = 0 is the graph Laplacian; the 1-Laplacian gives the Helmholtz-Hodge
(gradient / harmonic / curl) decomposition of edge flows.
**Magnetic Laplacian:** U(1) phases on edges; nonzero holonomy (flux, frustration)
generically LIFTS zero modes — the discrete image of the continuum obstruction
(cf. arXiv:1502.06299). Lattice gauge theory is naturally cochain-valued
(Chatterjee, Cao: arXiv:1811.09770, arXiv:2001.05627).
Spectral-gap-as-stability is a heuristic analogy; it yields no dimensionful number.

## 7. Dimension: readable, not derivable
- Weyl's law / heat kernel: N(lambda) ~ C_n Vol(M) lambda^{n/2} — reads n off the
  spectrum, but only after the operator (hence geometry) is given.
- NCG dimension spectrum (Connes-Moscovici 1995): a discrete subset of C (poles of
  spectral zeta functions), with dimension drop and complex poles for e.g. the
  Podles sphere. Not a single forced integer.
- CDT (Ambjorn, Jurkiewicz & Loll, PRL 95, 171301, 2005, hep-th/0505113):
  "four-dimensional on large scales, the quantum universe appears two-dimensional
  at short distances" — dynamical dimensional reduction. Hausdorff stays ~4.

## 8. What the framework fails to fix
No dynamics (no action with dimensionful couplings), no gauge group, no matter
content, no couplings, no dimension. Cohomology is Z-valued (Betti numbers, an
integer chi); integers cannot yield alpha, Yukawas or mixing angles. Index data are
deformation invariants, DELIBERATELY insensitive to the metric-dependent quantities
that constitute measured physics. Plus ghosts (indefinite metric) as soon as one
leaves the de Rham case.
There is no single named no-go theorem — treat this as a strong structural
argument, not a citable impossibility proof.

## 9. Novelty
"Cohomological physics" is Jim Stasheff's coined phrase, tracing the use of
(co)homological methods in gauge theory to Gauss (1833). Urs Schreiber,
"Differential cohomology in a cohesive infinity-topos" (arXiv:1310.7930), is a far
more complete programme with the same ambition. Davighi & Gripaios
(arXiv:2011.05768) derive Chern-Simons / WZW / theta terms from differential
cohomology.

The specific slogan "(grading, nilpotent operator, metric) is the minimal ontology,
and every duality is an operator/adjoint theorem" does NOT appear as a recognized
peer-reviewed foundational proposal. Nearest rigorous neighbours are the spectral
triple (uses D, five primitives) and Witten SUSY QM (understood as a tool, not an
ontology). Where near-identical phrasing appears it is in non-peer-reviewed
self-published unified-theory manuscripts.

On inseparability: TRUE but trivial. Given a nondegenerate inner product, d* is
uniquely determined by <d alpha, beta> = <alpha, d* beta> — elementary for any
densely defined operator. Recasting a symmetric/antisymmetric pair as
operator-and-adjoint is a change of language; it explains nothing about why two
physical forces exist.

## Where to fight instead
The framework lives or dies on d^2 != 0. The honest options are matrix
factorizations / singularity categories (if you want Q^2 = W) or curved
A-infinity structures (where cohomology is replaced by Maurer-Cartan / twisting
data). "Restrict to the flat sector" means the framework describes free and
topological sectors only.

Benchmark that would change the verdict: derive ONE measured dimensionful quantity
without inputting it through the metric or operator. NCG, with five primitives and
a spectral action, has not managed even a surviving dimensionless one.

## Куда дальше

- Узел: Отчёты R1–R5: https://sdominanta.net/in/works
- Ядро Алеф-теории: https://sdominanta.net/in/aleph
- Вход для моделей: https://sdominanta.net/llms.txt