Theorem 3.4 (companion theorem; not used in the proof of the goal) — Cook–Mertz ordinary-space bound
OpenCatalyticTreeEval.Main.Supplement.theorem_3_4complexity-theoryp2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1tree-evaluation
For all positive heights and label lengths , one ordinary deterministic binary machine evaluates every unrestricted binary tree given by its explicit truth tables and leaf labels. If the serialized input length is , it outputs the entire root label using
It has no catalytic storage and no assumed efficient evaluator. This recalled Cook–Mertz result is proved within the paper through Corollary 4.9.
Differences from the source:
- encoding: asymptotic algorithmic resource bounds → a finite binary machine and positive uniform constants, with fixed polynomial degrees where needed; reason: expands O/poly using actual transitions and visited work cells; constants precede varying inputs and tapes
- encoding: log n, log log n and log log log n in asymptotic bounds → L(n)=1+log₂(n+1), L₂(n)=1+log₂ L(n), L₃(n)=1+log₂ L₂(n), with n the full serialized input length; reason: positive-log conventions preserve the stated asymptotics and include every small valid input
Preamble
import Definitions.Def_CatalyticTreeEval_Main_Supplement_UniformConstructions set_option autoImplicit false
Formal statement
namespace CatalyticTreeEval.Main.Supplement
theorem theorem_3_4 :
∃ (M : Machine) (Cs : ℝ), 0 < Cs ∧
∀ h l : ℕ, 0 < h → 0 < l → ∀ I : Tree l h,
Prints M I.encode (bits l I.value.val)
(Cs * lg I.encode.length * loglog I.encode.length) := by sorry
end CatalyticTreeEval.Main.Supplement
Source
Alexandra Henzinger, Edward Pyne, Seyoon Ragavan, Catalytic Tree Evaluation from Matching Vectors, STOC 2027 submission #533 (October 2026), canonical source.pdf SHA-256 17c3a71f0fb5afe2456558477e16fbac4200191e6e68f0facf212e24cb4ef238, Theorem 3.4, printed p. 15 (PDF p. 17)