Binary de Bruijn Sequences via Zech's Logarithms

TL;DR

This paper presents a novel method combining Zech's logarithms with cycle joining techniques to efficiently construct binary de Bruijn sequences of any order, improving the process's speed and scalability.

Contribution

It introduces a new characterization of conjugate pairs via Zech's logarithms, enabling faster construction of de Bruijn sequences and handling large orders effectively.

Findings

Characterization of conjugate pairs using Zech's logarithms.

Efficient construction of connected adjacency subgraphs.

Scalable methods for large order de Bruijn sequences.

Abstract

The focus of this work is to show how to combine Zech's logarithms and each of the cycle joining and cross-join pairing methods to construct binary de Bruijn sequences of any order. A basic implementation is supplied as a proof-of-concept. The cycles, in the cycle joining method, are typically generated by a linear feedback shift register. We prove a crucial characterization that determining Zech's logarithms is equivalent to identifying conjugate pairs shared by any two distinct cycles. This speeds up the task of building a connected adjacency subgraph that contains all vertices of the complete adjacency graph. Distinct spanning trees in either graph correspond to cyclically inequivalent de Bruijn sequences. As the cycles are being joined, guided by the conjugate pairs, we track the changes in the feedback function. Certificates of star or almost-star spanning trees conveniently handle large order cases. The characterization of conjugate pairs via Zech's logarithms, as positional markings, is then adapted to identify cross-join pairs. A modified $m$-sequence is initially used, for ease of generation. The process can be repeated on each of the resulting de Bruijn sequences. We show how to integrate an analytic tool, attributed to Fryers, in the process. Most prior constructions in the literature measure the complexity of the corresponding bit-by-bit algorithms. Our approach is different. We aim first to build a connected adjacency subgraph that is certified to contain all of the cycles as vertices. The ingredients are computed just once and concisely stored. Simple strategies are offered to keep the complexities low as the order grows.