[Paper Review] A Computational Complexity-Theoretic Elaboration of Weak Truth-Table Reducibility

Kohtaro Tadaki|arXiv (Cornell University)|Jul 19, 2011

Computability, Logic, AI Algorithms23 references3 citations

TL;DR

This paper introduces a computational complexity-theoretic refinement of weak truth-table reducibility by explicitly bounding the query size via a function f, enabling asymptotic analysis within computability theory. It reveals a critical phase transition: while Z(T) and domU are weak truth-table equivalent for all computable T ∈ (0,1], only for T < 1 is domU reducible to Z(T) in query size O(n/T), revealing a fundamental asymmetry at T=1 due to the failure of the Ample Excess Lemma.

ABSTRACT

The notion of weak truth-table reducibility plays an important role in recursion theory. In this paper, we introduce an elaboration of this notion, where a computable bound on the use function is explicitly specified. This elaboration enables us to deal with the notion of asymptotic behavior in a manner like in computational complexity theory, while staying in computability theory. We apply the elaboration to sets which appear in the statistical mechanical interpretation of algorithmic information theory. We demonstrate the power of the elaboration by revealing a critical phenomenon, i.e., a phase transition, in the statistical mechanical interpretation, which cannot be captured by the original notion of weak truth-table reducibility.

Motivation & Objective

• To extend weak truth-table reducibility by explicitly bounding the query size using a computable function f, enabling complexity-theoretic analysis within computability theory.
• To investigate the computational relationship between the halting set domU and the generalized Chaitin Ω number Z(T) in algorithmic information theory.
• To identify a critical phase transition in reducibility behavior between T=1 and T<1 that is invisible under standard weak truth-table reducibility.
• To clarify why the Ample Excess Lemma fails to support unidirectional reduction from Z(1) to domU, unlike for T<1.

Proposed method

• Introduces 'reducibility in query size f' where an oracle Turing machine for A ⊆ {0,1}* only queries strings of length ≤ f(|x|) on input x.
• Applies the framework to the statistical mechanical interpretation of algorithmic information theory, focusing on the partition function Z(T) = ∑_{p∈domU} 2^(-|p|/T).
• Uses the program-size complexity H(α↾n) to analyze the partial randomness of Z(T), showing H(Z(T)↾n) ≤ Tn + O(1) for computable T ∈ (0,1].
• Employs the Ample Excess Lemma (for T=1) and Lemma 5.3(i) (for T<1) to establish bounds on the number of queries needed to recover domU from Z(T).
• Constructs a reduction from domF to Pf(Z_V(T)) using queries of size ⌈n/T⌉ + O(1), proving bidirectional reducibility for T<1.
• Demonstrates that for T=1, no such O(n/T)-query-size reduction exists from domU to Z(1), due to the failure of the weak Chaitin T-randomness condition.

Experimental results

Research questions

• RQ1 Does the standard weak truth-table reducibility distinguish between the computational behavior of Z(T) and domU at T=1 versus T<1?
• RQ2 Can a query-size-bounded reducibility framework reveal a phase transition in the reducibility between Z(T) and domU that is invisible to classical weak truth-table reducibility?
• RQ3 Why does the Ample Excess Lemma fail to support a unidirectional reduction from Z(1) to domU, while it enables such a reduction for T<1?
• RQ4 Is there a computable function f such that domU is reducible to Z(T) in query size f(n) for T<1 but not for T=1?
• RQ5 What role does the partial randomness level T play in determining the query-size complexity of reductions between Z(T) and domU?

Key findings

• For any computable T ∈ (0,1), domU is reducible to Z(T) in query size O(n/T), demonstrating a bidirectional relationship under the new framework.
• At T=1, no such O(n/T)-query-size reduction exists from domU to Z(1), revealing a critical phase transition at T=1.
• The Ample Excess Lemma, which enables unidirectional reduction from Ω to domU at T=1, fails to support the same for T<1 due to the breakdown of the weak Chaitin T-randomness condition.
• For T<1, Lemma 5.3(i) enables bidirectional reducibility between Z(T) and domU, but this lemma does not hold for T=1.
• The function f(n) = O(n/T) is optimal for reducing domF to Pf(Z_V(T)) when T<1, as shown by a contradiction in the growth rate of the program-size complexity.
• The paper establishes that the reducibility behavior between Z(T) and domU is fundamentally different at T=1 compared to T<1, with the latter allowing efficient query-size-bounded reductions.