On this page
Capped variance of random k-SAT: the statement
Imports 1 library modules and 7 external
External: HB.structures, mathcomp.classical.all_ssreflect_compat, mathcomp.algebra.algebraic_hierarchy.ssralg, mathcomp.algebra.numeric_hierarchy.ssrnum, mathcomp.reals.reals, mathcomp.analysis.sequences, mathcomp.analysis.exp
Imported by 1
Target T7 of phase 12 (AGENTS.md, section Phase 12; the plan notes/phase12-plan.md, section 3). The statement of a Rocq formalization of the credited argument of the proposition 'Polynomial concentration' of 'A limiting satisfiability threshold for every fixed clause size' (family 235 of the OpenAI math collection, revision 1581d4cbd, source/sections/concentration.tex), with the lemmas before it and the proposition of source/sections/forcing.tex. This file holds the statement only; its definitions are those of SatVarianceDefinitions.v. The Lean counterpart, in lean/OAI/Probability/RandomSAT of the collection:
Lean: variance_capped (Concentration.lean, line 80)
Setting. R is any realType and k >= 3. V_n is the length of the longest satisfiable prefix, capped at L_k n = 2^(k+1) n, of a tuple of L_k n independent uniform proper signed k-clauses on n variables, and sat_capped_var n k is its variance (SatVarianceDefinitions.v).
sat_variance_statement, for every R and every k >= 3: there is C > 0 such that Var(V_n) <= C n^(1 + 2/k) for every n >= k + 1, the power being powR.
Choices, for the reviewer.
- The manuscript writes Var(V_n) = O_k(n^(1 + 2/k)); here the constant C depends on k (and formally on R) only, and the bound holds for every n >= k + 1, the range of the Lean theorem, not only for large n.
- V_n is defined directly as the largest satisfiable prefix length up to the cap, on exactly L_k n clauses, as the user decided: it is min(H_n - 1, L_k n), with H_n the first unsatisfiable index, without an extra clause, and it covers the tuples whose every prefix is satisfiable.
- The Lean theorem bounds the variance of a lifetime defined by a recursion over the clause process, with the constant varianceConstant k n (n^(1/k))^2; the statement here is the manuscript's.
Declarations1 exported
PropDefinition sat_variance_statement : Prop := forall (R : realType) (k : nat), (3 <= k)%N -> exists C : R, 0 < C /\ forall n : nat, (k.+1 <= n)%N -> sat_capped_var R n k <= C * n%:R `^ (1 + 2 / k%:R).
Used by 1 declaration in 1 file
- GaussianApplications.SatVarianceTheorem: sat_variance_theorem