Reproduce “Connecting Network Science and Information Theory” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Henrique F. de Arruda, Filipi N. Silva, Cesar H. Comin, Diego R. Amancio, Luciano da F. Costa

Researchers at the University of São Paulo introduce a unified framework systematically integrating information theory and network science, demonstrating how an underlying network's topology and dynamics dictate the compressibility and transmissibility of generated information sequences. The framework shows that predicting symbol probabilities from network topology can yield better compression ratios, particularly when using dynamics like the True Self-Avoiding Random Walk for network sampling.

Key insights

  • The interaction between network topology and sampling dynamics fundamentally determines the compressibility and predictability of the generated information sequences.
  • Predicting symbol probabilities for Huffman coding from the underlying network's topology (e.g., node degrees) generally yields better compression ratios than using empirically observed symbol frequencies from the transmitted message.
  • The steering coefficient, which measures how much a dynamics is influenced by node degrees, is a valuable metric for understanding how network structure affects information generation and transmission.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.