Classification of countable structures asks whether objects have the same structure and whether one problem can supply definable invariants for another. Borel reducibility makes this comparison precise: a Borel map must preserve and reflect isomorphism. A class is Borel complete when every countable-structure isomorphism problem reduces to it. Friedman and Stanley introduced a hierarchy of bounded-rank tree problems and asked whether receiving every countable level already forces completeness. The question is stated in their original paper, page 899. Friedman–Stanley, 1989.
The relevant chronology distinguishes that general question from particular examples:
A countable relational language has at most countably many symbols, each with finite arity. A structure on the natural numbers assigns a truth value to each symbol and each tuple of the appropriate length. Nullary symbols are allowed. The resulting countable product of two-point spaces has its canonical Borel sigma algebra. Two structures are isomorphic when a single permutation of the natural numbers preserves all relation values.
An invariant Borel class is a Borel subset of that structure space closed under isomorphism. Borelness of the class does not assert Borelness of its isomorphism relation. Graphs are simple undirected graphs on the fixed vertex set of natural numbers.
For relations and , the notation means that some measurable map between their specified Borel code spaces satisfies
for every pair of source points. Neither relation is assumed Borel merely by writing this comparison.
The benchmark is equality on Cantor space . At successors, compares enumerated countable sets of classes, with a distinct empty-set code; order and repetitions are ignored. At nonzero countable limits, the benchmark is a coordinatewise product along a fixed increasing cofinal sequence of smaller ordinals. Codes remain actual measurable sequence spaces, rather than quotient spaces of invariants.
The goal is Theorem 1.2 of the supplied manuscript, not Question 1.1, which shares its numbering counter. There should be one fixed countable language and one invariant Borel class such that
and, including the theorem's additional conclusion,
Here denotes bounded-rank rooted well-founded tree codes, including finite and empty trees. The manuscript compares its product-limit benchmarks with the usual hierarchy conventions in Proposition 4.13 and with the original tree hierarchy in Proposition 4.15. No identification of ordinal indices between different presentations is asserted.
The milestones retain the source's attack path: the classical Borelness of bounded-rank tree isomorphism from Friedman–Stanley Section 1.2.1; Proposition 3.6 on invariant real-valued graph parameters; Theorem 4.1 on cofinality of Polishable subgroup coset relations; and Proposition 4.15 comparing the original tree hierarchy with the product-limit tower. The first is published background; the other three are targets from the proposed research draft.
The target distinguishes being above every countable benchmark from being universal for all countable-structure classification problems. In the source's sentence formulation, it would give a negative answer to the question restated by Shelah and Ulrich. The structure-class formulation is the goal here; the infinitary-sentence corollary and the appendix's forcing and thickness claims are outside this mission.
The requested formalization is open. Local compilation validates the definitions and theorem interfaces, with explicit open-target placeholders; it does not verify the proposed mathematical proof. A completed development would provide reusable code-space constructions, ordinal-indexed comparison relations, bounded-tree interfaces, and permutation-group coset machinery.
The quantifier ranges over every countable ordinal, while the language and target class must remain fixed. Results restricted to finitely many jumps, or even to the omega level, do not settle this claim. The distinction between countable sets and countable multisets is also substantive: the original tree codes retain multiplicities, while the set jump forgets them. Friedman–Stanley, Section 1.2.
The class of structures must itself be Borel and invariant. Replacing that requirement by Borelness of its isomorphism graph changes the problem. Individual reductions may depend on the ordinal; the theorem does not require a single Borel selection of reductions over all codes for countable well-orders.
The Lean language bundle represents actual relational structures on by Boolean truth tables. The source allows functions and constants to be represented by relations; no structural axioms are imposed on the ambient space. A proposed target class must supply its own Borelness and invariance. Empty languages and empty classes are admitted by the interfaces, and no nontriviality is hidden in those definitions.
The tower uses native ordinal recursion below , equality on Cantor space, option-valued sequence codes for the set jump, and dependent countable products at limits. Fundamental sequences are fixed once; their strict increase and cofinality in the countable range are proved in the definition layer. This commits to the manuscript's product-limit presentation.
Tree codes are prefix-closed subsets of finite natural-number sequences. Isomorphisms are actual node bijections preserving and reflecting immediate parenthood. A global bound is expressed by an ordinal ranking into the bound, strictly decreasing along children; the empty tree has height zero and a single vertex height one. This specifies reverse-tree height explicitly and does not conflate the original convention with the separate leaf-zero padding convention. Finite and infinite branch multiplicities are retained.
The coset interface uses the actual group of permutations of with its standard Polish topology and the source convention . No normality or closedness of the image subgroup is assumed. Contributions developing the ordinal, measurable-space, tree, and topological-group prerequisites are welcome.
FriedmanStanleyQuestion.tex, Theorem 1.2; Proposition 3.6; Theorem 4.1; Propositions 4.13 and 4.15. Author field empty; no public manuscript URL supplied.theorem FSCofinality.theorem1_2 :
∃ L : FSCofinality.CountableRelationalLanguage,
∃ K : Set (FSCofinality.Structures L),
MeasurableSet K ∧ FSCofinality.IsInvariant K ∧
(∀ a : Ordinal.{0}, a < Ordinal.omega 1 →
(FSCofinality.fsTower a).Reduces (FSCofinality.classRelation L K)) ∧
¬ FSCofinality.graphRelation.Reduces (FSCofinality.classRelation L K) ∧
(∀ a : Ordinal.{0}, 0 < a → a < Ordinal.omega 1 →
(FSCofinality.RootedTrees.treeRelation a).Reduces
(FSCofinality.classRelation L K)) := by sorry
There are one fixed countable relational language L and one isomorphism-invariant Borel class K of L-structures on the natural numbers such that every countable level of the concrete product-limit Friedman--Stanley hierarchy Borel reduces to isomorphism on K, while graph isomorphism does not. Moreover, bounded-rank well-founded rooted tree isomorphism reduces to the same target for every nonzero countable bound, including finite and empty tree codes. Borelness is a property of K as a class of structures, not a requirement that its isomorphism graph be Borel. The language and target class are fixed before the ordinal quantifiers, while each ordinal may have its own reduction. The product-limit and original-tree presentations make the cofinality convention explicit.
The same and satisfy
together with
Here uses the explicitly defined global reverse-tree-height presentation; ordinal indices are not identified with those of a different rank convention.
No open leaves. Every sub-goal is proved or awaiting decomposition.