Generalized and Unified Equivalences between Hardness and Pseudoentropy

TL;DR

This paper establishes a unified pseudoentropy characterization linking computational hardness and randomness, applicable to various entropy notions, and introduces technical tools that improve complexity bounds significantly.

Contribution

It generalizes pseudoentropy characterizations to a broad family of entropy notions using weight-restricted calibration, unifying and strengthening previous results.

Findings

Unified pseudoentropy characterization for multiple entropy notions.

Single universal function witnesses computational hardness and randomness.

Exponential improvement in complexity dependency on alphabet size.

Abstract

Pseudoentropy characterizations provide a quantitatively precise demonstration of the close relationship between computational hardness and computational randomness. We prove a unified pseudoentropy characterization that generalizes and strengthens previous results for both uniform and non-uniform models of computation. Our characterization holds for a general family of entropy notions that encompasses the common notions of Shannon entropy and min entropy as special cases. Moreover, we show that the characterizations for different entropy notions can be simultaneously achieved by a single, universal function that simultaneously witnesses computational hardness and computational randomness. A key technical insight of our work is that the notion of weight-restricted calibration from the recent literature on algorithm fairness, along with standard computational indistinguishability (known as multiaccuracy in the fairness literature), suffices for proving pseudoentropy characterizations for general entropy notions. This demonstrates the power of weight-restricted calibration to enhance the classic Complexity-Theoretic Regularity Lemma (Trevisan, Tulsiani, and Vadhan, 2009) and Leakage Simulation Lemma (Jetchev and Pietrzak, 2014) and allows us to achieve an exponential improvement in the complexity dependency on the alphabet size compared to the pseudoentropy characterizations by Casacuberta, Dwork, and Vadhan (2024) based on the much stronger notion of multicalibration. We show that the exponential dependency on the alphabet size is inevitable for multicalibration as well as for the weaker notion of calibrated multiaccuracy.