Consolidated Results Through Run 28

Result map

Run Native model Exact advance Principal obstruction retained
1 Bottom-up tree automata Contextual equivalence is the largest reachable finality-saturating congruence; distinguishing depth is sharply at most (n-2) Unreachable constants create fictional contexts
2 Fixed-order 3CNF / ROBDD (2^k) histories collapse to two semantic residuals in the baseline family Syntactic residual counts depend on provenance conventions
3 Ordered SAT prefixes Future equivalence is the greatest graded congruence Erasing prefix grade makes next-bit update ill-defined
4 Uniform streaming Positional tasks need exactly (j+2) uniform residual states despite cut width two Treating the clock as free hides state cost
5 Rewriting aa→aaa\to a Raw histories count ((n-1)!); square classes count (C_{n-1}) Confluence does not identify overlapping histories
6 Four-leaf coherence Schedule quotient is an unfilled pentagon with π1≅Z\pi_1\cong\mathbb Z Connectedness is weaker than coherence
7 Five-leaf coherence Six pentagons and three squares form a complex homotopy equivalent to (S^2) Homology alone would not prove simple connectedness
8 Dynamic tree cache Disjoint fixed replacements commute; batch saving is exactly LCA depth plus one Cost is algorithm-relative
9 Early stopping Commuting updates can have sharp order gap (2h) Semantic commutation does not imply resource flatness
10 Ancestor chains Nested kernels induce an ultrametric; cross-level cost biases cannot cancel Mutable contexts can break nestedness
11 Arbitrary incomparable updates Work factors through branch costs and the LCA corner square Spatial data alone omits the LCA transition algebra
12 Persistent DAG copying Immutable occurrence-path copying is alias-transparent Destructive identity and canonical reuse cross the boundary
13 Hash-consed sibling updates Allocation gap is at most one and depends on retained intermediate roots Equal cost can leave unequal tables
14 Height-two hash-consing Common-request coverage replaces naive initial membership Cross-level identity creates endogenous prepayment
15 Arbitrary hash-cons spine Exact set-rank law; sharp gap (m); chain biases cannot cancel Unrestricted key-set cancellation is not structurally realizable
16 Dependency posets Ideal lattices exactly classify cancellation; sign coherence iff the poset is a chain Native single-path copying cannot generate a fork
17 Learned 3CNF resolvent Clause-minimal semantic/UP/decision separation Learned binary receipt and accounting policy are explicit
18 SAT products DPLL/UP decision cost is additive over disconnected clause components Semantic product structure alone does not imply additivity
19 Connected resolution chain Pure-resolution receipts give exact connected cost (k- S
20 Receipt amortization Exact acquisition/reuse frontier (r−1)d≥c+s(r-1)d\ge c+s Deletion destroys amortization
21 Demand coverage Break-even depends exactly on covered future demands Uncovered demands do not repay acquisition
22 Proof-state fiber / OBDD Exact hierarchy 2k→k+1→12^k\to k+1\to1 for proof state, scalar cost, semantics Full closure or deletion changes the fiber
23 Propagation demands Exact information law: demands (T) require ( T
24 Second-order branching Same singleton UP signature, different fixed-policy DPLL cost Singleton tests are vacuous for strict 3CNF
25 Arbitrary clause width Width (r) first differs at interaction depth (r-1) Unbounded depth is obtained by increasing width
26 Fixed-width absorption Seed depth (w-1) exactly characterizes all-seed UP equivalence Fixed-width UP cannot support unbounded escape depth
27 First-UIP provenance All-seed UP-equivalent databases have an unbounded selected-trace pivot gap Clause priority and reason selection matter
28 All-schedule first-UIP Base spectrum is ({m+1}), augmented spectrum ({1,m+1}); optimum gap (m) A shortcut enables but does not force lower online cost

Current frontier

The program has established a stable hierarchy of distinctions:

same semantic endpoint⇏same operational history⇏same retained representation⇏same resource cost.\text{same semantic endpoint} \not\Rightarrow \text{same operational history} \not\Rightarrow \text{same retained representation} \not\Rightarrow \text{same resource cost}.

The tree campaign exposed contextual congruences, LCA geometry, ultrametric propagation, and set-rank growth under canonical sharing. Its arbitrary-height spine and ideal-lattice theorems closed the height-by-height line and forced a pivot rather than further enumeration.

The SAT campaign then separated semantic residuals, clause-database state, unit-propagation readiness, decision cost, and reason provenance. The latest pair (F_m,G_m) has identical semantics and identical all-seed UP closure, yet retaining one absorbed resolvent changes the minimum legal first-UIP analysis from (m+1) pivots to one. Run 28 proves this is not an artifact of one fixed priority, while also showing that the shortcut does not improve the worst legal schedule.

Current obligation

The reason-path family is classified under fixed and arbitrary schedules. The next bounded SAT/proof target should test a resource not reducible to that scheduling spectrum: the smallest connected strict-3CNF pair with identical semantics and all-seed UP closure but different minimum resolution width or clause space for a fixed unsatisfiable residual, or a precise obstruction.