The Quaternary Gray Code and How It Can Be Used to Solve Ziggurat and Other Ziggu Puzzles

TL;DR

This paper introduces the Ziggu family of puzzles, encodes their states using quaternary Gray code, and provides efficient algorithms for solving, ranking, and comparing puzzle states, expanding understanding of Gray code applications.

Contribution

It presents the first analysis of Ziggu puzzles using quaternary Gray code, including solution algorithms and a bijection to Nurikabe grids, advancing puzzle encoding methods.

Findings

Shortest Ziggu solution length is $6 imes 2^n - 3n - 5$

Longest Ziggu solution visits all states, following the quaternary Gray code order

Provides $O(n)$ algorithms for ranking, comparison, and successor operations

Abstract

We investigate solutions to the new "Ziggu" family of exponential puzzles. These puzzles have $p$ pieces that form $m$ mazes. We encode the puzzle state as an quaternary number (base $4$) with $n=m+1$ digits, where each digit gives the horizontal or vertical position in one maze. We show that the number of states on a shortest solution is $6 \cdot 2^n - 3n - 5$ (OEIS A101946). This solution is unique, and it is generated from the start as follows: make the leftmost modification that doesn't undo the previous modification. Replacing "leftmost" with "rightmost" instead generates the unique longest solution that visits all $(3^{n+1} - 1)/2$ states (OEIS A003462). Thus, Ziggu puzzles can be viewed as $4$-ary, $3$-ary, or $2$-ary puzzles based on how the number of state encodings, valid states, or minimum states grow with each additional maze. Classic Gray code puzzles (e.g., Spin-Out) provide natural comparisons. Gray code puzzles with $p$ pieces have $2^p$ (OEIS A000079) or $\lfloor \frac{2}{3} \cdot 2^p \rfloor$ (OEIS A000975) states on their unique solution. The states visited in a Gray code puzzle solution follow the binary reflected Gray code. We show that Ziggu puzzles follow the quaternary reflected Gray code, as the shortest and longest solutions are both sublists of this order. These results show how to solve Ziggu puzzles from the start. We also provide $O(n)$-time ranking, comparison, and successor algorithms, which give the state's position along a solution, the relative order of two states, and the next state, respectively. While Gray code puzzles have simpler recursive descriptions and successor rules, the Ziggu puzzle has a simpler loopless algorithm to generate its shortest solution. The two families share a comparison function. Finally, we enrich the literature on OEIS A101946 by providing a bijection between Ziggu states and $2\times n$ Nurikabe grids.