What geometry supplies the polarization?

Run status: not attempted — no run in lab/runs/ names this case in its settles: field.

Object. The adele class space with its scaling flow, as a candidate home for a Lefschetz-trace proof of the Riemann hypothesis. The trace formula is in hand and verified place by place: one local integral per place, of the Atiyah–Bott shape ∫|u|^{1/2}G(log|u|)/|1−u|_v d*u. At a finite place the identity |1−u|_p = max(1,|u|_p) for |u|_p ≠ 1 makes it exact rather than asymptotic (verified, no failures at p = 2,3,5,7), giving weight exactly p^{−k/2} with no (1−p^{−k}) correction. The archimedean term agrees with the digamma form up to one universal constant, γ + log 2π, across six test functions (spread 0.002).

Presenting problem (the friction). Everything needed for a Lefschetz argument is present except the sign, and the sign is not a missing lemma — it is a missing object.

Over a finite field the analogous statement is a signature computation. On the surface C × C with h = f₁ + f₂, the intersection form in the basis (f₁, f₂, D, F) at q = 7, g = 1 has signature (1,3) for a = 2, −4, 5 and (2,2) for a = ±6, i.e. flips exactly when |a| exceeds 2√q; equivalently h² = 2 > 0 with negative-definite orthogonal complement (top eigenvalue −0.033 at a = 5, +0.101 at a = 6). Positivity there means: an ample class with negative-definite complement, which is the Lefschetz decomposition.

The adele class space has a pairing, and a trace with real-valued dimensions. It has no ample class and no hard-Lefschetz sl₂. And this cannot be repaired algebraically: over ℂ all nondegenerate symmetric forms are equivalent, signatures exist only over ℝ, and a real form is not determined by its complexification. An ample cone is a cone in a real vector space. So positivity is not derivable from the algebra; it has to be imported from a geometry, and the question is which one.

Differential. 1. The missing geometry is a Riemann–Roch over the field with one element. The ample class and the sl₂ come with the intersection theory once the base object exists. 2. The missing geometry is a foliated space, with the flow as the leafwise dynamics and positivity from a leafwise Hodge theory. 3. The ample class exists but on a different object than anyone has been looking at — not the space, but the span of the correspondences Ψ(λ) themselves, where the demand is that this span be negative definite modulo fibres and diagonal. 4. No import is possible, because the positivity statement is equivalent to the target and any geometry rich enough to supply it already contains the conclusion. The programme is then circular in a way that is not obvious from inside it.

Defect the panel found (PN-001). The fourth branch is not symmetric with the other three. "No import possible" is not a rival mechanism — it is this case's own verdict condition wearing a hypothesis's coat: the conclusion the case invites if the half-space cannot be named. The enumeration smuggles its own answer in as a branch, so the differential is not a clean partition of mechanisms; branch 4 should be recast as the case's negative outcome, not scored against branches 1–3.

Discriminating tests — House's seven (from PN-001).

1. Derive γ + log 2π from the digamma asymptotics by a route that does not assume the trace formula's normalisation. Cheapest, pure computation, runs first: a residual at the precision floor means it is an identity; no better than the existing 0.002, or no normalisation-independent route at all, means it is a fitted constant and the archimedean premise drops to provisional.
2. The decisive test, and it is one test not two. Attempt a conjugation-invariant splitting — reflected coordinate, one-sided subalgebra, ⟨θf,f⟩ on a finite truncation — at each of several archimedean spectral-parameter values, sweeping the parameter. No splitting at any value → account 4; splitting with sign constant across the sweep → positivity is an attribute of the space and branches 1–3 become rankable; splitting with sign flipping at a threshold → positivity is parameter-indexed and all four branches are malformed. Run parameterised or not at all — a null on existence alone is uninterpretable.
3. Is the adele-class pairing genuinely a complexification of a real form, or real-valued from the outset with nothing to descend from? If the latter, "a real form is not determined by its complexification" is true but irrelevant here and the friction's central move collapses.
4. Does the adelic object carry any finite-rank intersection-form analogue, or is the (1,3)/(2,2) flip an artefact of the surface's product structure h = f₁ + f₂? If the pairing is genuinely infinite-rank with real-valued dimensions, the case's strongest evidence does not transfer.
5. Test the load-bearing premise directly: does the verified trace formula, absent any positivity input, already constrain the spectral configuration? Off-line configurations admissible → positivity is a genuine separate input; inadmissible → the sign was never the missing object and the friction is misdiagnosed.
6. Construct a fifth account in which positivity is a spectral condition on the flow rather than a geometric structure. If one exists, the four-way enumeration was never well-posed.
7. State what would make the imported positivity checkable independently of the Lefschetz conclusion. Non-circular if it reduces to an already-verified identity; circular if every candidate check presupposes the conclusion — which would make branch 4 true by construction rather than by evidence.

Stakes. Whether "import a geometry" is a research instruction or a restatement of the problem. Branch 4 is the one that would hurt, and it is not currently excluded.

---

Update (2026-09-12): the measure problem is the same problem.

A second instance of this case's structure has surfaced in a setting with an eighty-year literature of its own. The path integral's existence is a positivity statement: there is no Lorentzian measure, the Euclidean measure exists exactly under reflection positivity, and Wick rotation is the choice of a real form.

That is this case's shape, verbatim. An algebraic object (the formal path integral, the trace formula) with everything present except a sign; a positivity condition that cannot be derived from the algebra; and an import from a real structure that makes the object exist. Osterwalder–Schrader is the successful instance of the import, and constructive field theory is the record of how hard the import is in practice.

Why this matters for the differential above. Branch 4 — that no import is possible because any geometry rich enough to supply the positivity already contains the conclusion — is weakened, and should be. Reflection positivity is a genuine imported condition that is checkable independently of what it is used to prove, in the cases where the construction succeeds. It is not a restatement of the target. So there is at least one worked example of the move this case is asking for, in which the import was real, non-circular, and extremely difficult.

A test this makes available. Examine whether the adele class space admits a reflection — an involution with respect to which a positivity condition could be stated — and what it would reflect. The criteria list already records conjugation as existing while positivity does not, which is the same gap in different words: a reflection without a positivity condition attached to it. Osterwalder–Schrader's content is not the existence of the reflection but the positivity of the pairing it induces on a half-space. The question is therefore what plays the role of the half-space, and the archimedean place's status as the singular fibre and the only boundary-like direction is the obvious candidate.

Revised stakes. Whether "import a geometry" is a research instruction rather than a restatement now has a precedent answer: in field theory it was an instruction, it took decades, and it worked in low dimensions and not in four.