Depth-Constrained Binary Prefix Trees: Asymmetric Optima and Rényi-Optimum Switching
This preprint studies binary prefix trees with a fixed number of leaves and a maximum depth. It examines how unequal branching probabilities affect the tree structures that minimize Rényi entropy for orders greater than one. The main finding is that the optimal structure can change as the Rényi order increases, even when the branching probabilities, leaf count, and depth limit remain fixed. The paper proves exact conditions for these changes in specific tree families, including an infinite family with exactly two changes of optimum. It also compares these results with known symmetric cases and characterizes when a flow relaxation based on mixtures of trees matches the tree optimum in the symmetric setting. This record contains the English preprint and a standalone numerical reproducibility package. The package includes Python source code, instructions, verified computational results, and tools for reproducing the numerical tables and figures. It requires Python 3.12 or later and uses only the standard library. The finite computational checks supplement the mathematical proofs presented in the paper. The paper and the reproducibility package are released under the Creative Commons Attribution 4.0 International licence (CC BY 4.0).
Authors
- Zsolt Kisander (ORCID: https://orcid.org/0000-0001-9245-4728)
Institutions
- University of Pecs (HU)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.23507968
- Primary Topic
- Algorithms and Data Compression
- Type
- preprint