Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
169 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A Decomposition Method for Global Evaluation of Shannon Entropy and Local Estimations of Algorithmic Complexity (1609.00110v7)

Published 1 Sep 2016 in cs.IT, cs.CC, and math.IT

Abstract: We investigate the properties of a Block Decomposition Method (BDM), which extends the power of a Coding Theorem Method (CTM) that approximates local estimations of algorithmic complexity based upon Solomonoff-Levin's theory of algorithmic probability providing a closer connection to algorithmic complexity than previous attempts based on statistical regularities e.g. as spotted by some popular lossless compression schemes. The strategy behind BDM is to find small computer programs that produce the components of a larger, decomposed object. The set of short computer programs can then be artfully arranged in sequence so as to produce the original object and to estimate an upper bound on the length of the shortest computer program that produces said original object. We show that the method provides efficient estimations of algorithmic complexity but that it performs like Shannon entropy when it loses accuracy. We estimate errors and study the behaviour of BDM for different boundary conditions, all of which are compared and assessed in detail. The measure may be adapted for use with more multi-dimensional objects than strings, objects such as arrays and tensors. To test the measure we demonstrate the power of CTM on low algorithmic-randomness objects that are assigned maximal entropy (e.g. $\pi$) but whose numerical approximations are closer to the theoretical low algorithmic-randomness expectation. We also test the measure on larger objects including dual, isomorphic and cospectral graphs for which we know that algorithmic randomness is low. We also release implementations of the methods in most major programming languages---Wolfram Language (Mathematica), Matlab, R, Perl, Python, Pascal, C++, and Haskell---and a free online algorithmic complexity calculator.

Citations (107)

Summary

  • The paper presents the Block Decomposition Method (BDM), a novel approach offering an improved approximation of algorithmic complexity by combining insights from Shannon entropy and local estimations.
  • BDM decomposes objects into smaller blocks, evaluates local algorithmic complexity for each, and sums these to estimate the complexity of the entire structure.
  • Numerical results demonstrate BDM's effectiveness in estimating complexity for large objects and recognizing algorithmically compressible sequences previously seen as random by entropy methods.

Analyzing Algorithmic Complexity through Block Decomposition

This paper presents a novel methodology termed the Block Decomposition Method (BDM) meant to enhance the understanding of algorithmic complexity by providing a better approximation to Kolmogorov complexity than traditional methods. Building upon the Coding Theorem Method (CTM), BDM offers an avenue to explore local estimations of algorithmic complexity while maintaining universal properties, unlike common statistical methods like Shannon entropy.

Overview of the Approach

At the core of the paper is the proposition that employing small computer programs, which produce segments of a larger object, can help approximate the object’s algorithmic complexity more efficiently. The methodology decomposes a given object into smaller blocks, evaluates each for algorithmic complexity, and combines these insights to infer the complexity of the entire structure. BDM quantifies complexity by summing up local estimations of the individual components of an object, effectively forming a bridge between classical Shannon entropy and algorithmic (Kolmogorov-Chaitin) complexity metrics.

Strong Numerical Results and Analysis

BDM, as proposed, manages to offer a feasible solution to estimating the algorithmic complexity for large objects, a task previously hindered by the non-computability of Kolmogorov complexity. It performs well also in numerical experiments, suggesting effective estimations of complexity even among objects with apparent randomness under traditional entropy measures. The authors demonstrate the capability of BDM to recognize algorithmically compressible sequences otherwise perceived as random.

For instance, the paper illustrates testing on known low-algorithmic-complexity sequences, such as the Thue-Morse sequence or digits of π, revealing that BDM assigns these sequences lower complexity than it would under a solely entropy-based evaluation.

Implications and Future Directions

The implications of this research span both theoretical advancements and practical applications. From a theoretical standpoint, the hybrid nature of BDM, which tends toward Shannon entropy in the worst-case scenario but captures algorithmic nuances otherwise, underscores its innovation when set against existing approximations of Kolmogorov complexity. Practically, this work equips researchers with a more refined tool for complexity analysis widely applicable in domains such as data compression, cryptography, and information theory.

As computational resources advance, future research may push the boundaries further by improving CTM precomputation for broader scales or adapting BDM in multidimensional contexts like tensors or graph structures. The compression performance, its enhancement through integrating CTM and BDM approaches, and the development of sophisticated algorithms indicate the evolving frontier of complexity analysis. This trajectory could potentially converge with contemporary AI approaches, considering algorithmic complexity in evaluating model architectures or as a metric for generative models.

In conclusion, this paper enriches the landscape of complexity theory, offering a methodological breakthrough that supplements both classical and algorithmic measures, thus reinforcing the coherence of complexity approximations in computational contexts.

Youtube Logo Streamline Icon: https://streamlinehq.com