Answers are grounded in this paper and cite their sources.
Local objects may exist everywhere and be compatible everywhere, while aglobal object may still fail to exist; this failure mode has long been covered bythe single phrase “failure of the Hasse principle.” The present paper formalizesthat mode as an auditable object: descent data carry a cochain, the class of thatcochain is the gluing genus, the nontriviality of a torsor is the geometric form of thegluing genus, and under these coordinates the problem of global closure becomesa purely algebraic decision. Three theorem-level statements are given: a descenteffectiveness criterion—descent data are effective if and only if the correspondingtorsor has a global section; a pointed-set theorem for non-abelian cohomology—when the structure group is non-abelian, first cohomology has in general no groupstructure, so the phrase “nonzero genus group” is legitimate only after passage toan abelianized level, and this distinction is the disciplinary basis for all subsequentnotation; and an abelian-level gluing theorem—for an abelian group sheaf, localpieces can be glued into a global section if and only if the gluing class vanishes.The obstruction ladder is registered as the inclusion chain X(k) ⊆ X(A)´etBr ⊆X(A)Br ⊆ X(A); whether equality holds in any link is a deep problem, and thispaper only registers the level at which an object dies, without predicting the endof the ladder. The type discipline is fixed throughout in two notational forms: atthe abelian level one writes GDab = 0 or GDab ̸ = 0, while at the general nonabelian level one writes [D] = ∗ or [D] ̸ = ∗; the triggering conditions for the twoforms are checked place by place against the general rule. In Section 8.4 the Selmercurve C : 3x3 + 4y3 +5z3 = 0 gives the first load-bearing arithmetic sample: itsclass [C] ∈ H1(Q,E) is a nontrivial three-torsion element of Sha(E/Q); it is locallysoluble everywhere and dies at the level “the gluing class is nonzero.” The threecolumn table (local objects; compatibility data; the missing coboundary) is herefilled for the first time by a load-bearing example. Four machine-check readingsare completed before the main text: a Hilbert-90 reconciliation over a quadraticextension of a finite field; a minimal example of a cochain on a three-open cover thatis not a coboundary; a toy ledger of exclusion by diagonal closure in the adeles; and ateaching anchor in which the scissors for a conic are empty. The mechanism sentenceof Reading Three is registered as a boxed conditional: it holds inside the boxconstraint |a| ≤ 200, while the outside-box counterexample q = 900/899 shows thatthe sentence is never used beyond the box boundary. Throughout, readings are onlyregistered and never promoted, and negative conclusions are always conditionalized.The four-condition converse upgrades [D] = ∗ to an effective construction; theexistential converse is given unconditionally by Theorem 1, and only the upgradeto effective construction can be conditionalized. The paper nowhere writes aninclusion chain as an equality chain, nowhere treats nonemptiness of a BrauerManin set as existence, and nowhere conflates the two Hilbert-tenth conclusionsover the ring of integers and over the field of rationals.
No takes yet. Share an insight, caveat, or question.
zhou et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: