Abstract:De Bruijn tori, also called perfect maps, are two-dimensional periodic arrays of letters drawn from a given finite alphabet, such that each possible pattern of a given shape $(m,n)$ appears exactly once within one period of the torus. It is still unknown if de Bruijn tori of some certain size exist, like e.g. square shaped de Bruijn Tori with odd $m=n\in\{3,5,7\}$ and an even alphabet size $k$. However, in certain applications like positional coding, sub-perfect maps are sufficient, i.e. one does not need every possible $(m,n)$-pattern to appear, as long as a sufficient large number of such patterns is captured and every pattern occurs at most once. We show, that given any $m=n$ and a square alphabet size $k^2$, one can efficiently construct a sub-perfect map which is almost perfect, i.e. of almost maximal size. We do this by introducing de Bruijn rings, i.e. sub-perfect maps of minimal height, and providing an efficient construction method for them. We extend our results to non-square torus shapes and arbitrary non-prime alphabet sizes.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: For de Bruijn tori of certain specific sizes, especially square de Bruijn tori (i.e., \(m = n\) and \(m, n\in\{3, 5, 7\}\)) and the case of even alphabet sizes, it is not clear whether such structures exist. In some application scenarios, such as position coding, sub - perfect maps are sufficient, that is, not every possible \((m, n)\)-pattern needs to appear, as long as enough patterns are included and each pattern appears at most once.
Specifically, the main objectives of the paper include:
1. **Constructing almost - perfect sub - perfect maps**: For a given \(m = n\) and square alphabet size \(k^{2}\), provide an efficient method to construct almost - perfect sub - perfect maps.
2. **Proving the near - perfection of these maps**: As \(m, n\) increase, the proportion of patterns that do not appear tends to zero.
3. **Extending to non - square tori and other non - prime alphabet sizes**: Generalize the results to non - square torus shapes and arbitrary non - prime alphabet sizes.
### Overview of the paper's solution method
To achieve the above objectives, the paper introduces the concept of **de Bruijn rings**, that is, sub - perfect maps with the minimum height. By introducing de Bruijn ring graphs and using Eulerian cycles for construction, the paper shows how to efficiently generate these sub - perfect maps.
#### Key steps
1. **Define de Bruijn rings**: Sub - perfect maps with height \(m\) and width \(M(k^{n}, m)\).
2. **Construct ring graphs**: Based on the fact that the edges of the ring graph correspond to unique row - Lyndon patterns, and find Eulerian cycles by the Hierholzer algorithm.
3. **Generate de Bruijn rings**: Gradually add columns through the algorithm, ensuring that the resulting map after each round of cycling is a row - Lyndon pattern.
4. **Combine multiple de Bruijn rings**: Generate larger - sized sub - perfect maps by combining de Bruijn rings of different alphabet sizes.
### Conclusion
By introducing de Bruijn rings and their efficient construction methods, the paper successfully solves the problem of the existence of de Bruijn tori of specific sizes and provides near - perfect sub - perfect maps, which are suitable for application scenarios that need to cover most patterns. In addition, the paper also proves that these maps almost contain all possible patterns at larger sizes, thus verifying their near - perfection.