Kazuya Haraguchi

Approximability of Latin Square Completion-Type Puzzles

Among many variations of pencil puzzles, Latin square Completion-Type puzzles (LSCP), such as Sudoku, Futoshiki and BlockSum, are quite popular for puzzle fans. Concerning these puzzles, the solvability has been investigated from the viewpoint of time complexity in the last decade; it has been shown that, in most of these puzzles, it is NP-complete to determine whether a given puzzle instance has a proper solution. In this paper, we investigate the approximability of LSCP. We formulate LSCP as the maximization problem that asks to fill as many cells as possible, under the Latin square condition and the inherent condition. We then propose simple generic approximation algorithms for LSCP and analyze their approximation ratios.

An Efficient Local Search for Partial Latin Square Extension Problem

A partial Latin square (PLS) is a partial assignment of n symbols to an nxn grid such that, in each row and in each column, each symbol appears at most once. The partial Latin square extension problem is an NP-hard problem that asks for a largest extension of a given PLS. In this paper we propose an efficient local search for this problem. We focus on the local search such that the neighborhood is defined by (p,q)-swap, i.e., removing exactly p symbols and then assigning symbols to at most q empty cells. For p in {1,2,3}, our neighborhood search algorithm finds an improved solution or concludes that no such solution exists in O(n^{p+1}) time. We also propose a novel swap operation, Trellis-swap, which is a generalization of (1,q)-swap and (2,q)-swap. Our Trellis-neighborhood search algorithm takes O(n^{3.5}) time to do the same thing. Using these neighborhood search algorithms, we design a prototype iterated local search algorithm and show its effectiveness in comparison with state-of-the-art optimization solvers such as IBM ILOG CPLEX and LocalSolver.

A Constructive Algorithm for Partial Latin Square Extension Problem that Solves Hardest Instances Effectively

A partial Latin square (PLS) is a partial assignment of \(n\) symbols to an \(n\times n\) grid such that, in each row and in each column, each symbol appears at most once. The partial Latin square extension (PLSE) problem asks to find such a PLS that is a maximum extension of a given PLS. Recently Haraguchi et al. proposed a heuristic algorithm for the PLSE problem. In this paper, we present its effectiveness especially for the “hardest” instances. We show by empirical studies that, when \(n\) is large to some extent, the instances such that symbols are given in 60–70 % of the \(n^2\) cells are the hardest. For such instances, the algorithm delivers a better solution quickly than IBM ILOG CPLEX, a state-of-the-art optimization solver, that is given a longer time limit. It also outperforms surrogate constraint based heuristics that are originally developed for the maximum independent set problem.

On a generalization of Eight Blocks to Madness puzzle

We consider a puzzle such that a set of colored cubes is given as an instance. Each cube has unit length on each edge and its surface is colored so that what we call the Surface Color Condition is satisfied. Given a palette of six colors, the condition requires that each face should have exactly one color and all faces should have different colors from each other. The puzzle asks to compose a 2 × 2 × 2 cube that satisfies the Surface Color Condition from eight suitable cubes in the instance. Note that cubes and solutions have 30 varieties respectively. In this paper, we give answers to three problems on the puzzle: (i) For every subset of the 30 solutions, is there an instance that has the subset exactly as its solution set? (ii) Create a maximum sized infeasible instance (i.e., one having no solution). (iii) Create a minimum sized universal instance (i.e., one having all 30 solutions). We solve the problems with the help of a computer search. We show that the answer to (i) is no. For (ii) and (iii), we show examples of the required instances, where their sizes are 23 and 12, respectively. The answer to (ii) solves one of the open problems that were raised in [E. Berkove et al., “An Analysis of the (Colored Cubes) 3 Puzzle,” Discrete Mathematics, 308 (2008) pp. 1033–1045].

How Simple Algorithms Can Solve Latin Square Completion-Type Puzzles Approximately

Among many variations of pencil puzzles, Latin square Completion-Type puzzles (LSCPs) are quite pop- ular for puzzle fans. Concerning these puzzles, the solvability has been investigated from the viewpoint of time com- plexity in the last decade; it has been shown that, in most of these puzzles, it is NP-complete to determine whether a given puzzle instance has a proper solution. In this paper, we investigate the approximability of three LSCPs: Sudoku, Futoshiki and KenKen. We formulate each LSCP as a maximization problem that asks to fill as many cells as possible, under the Latin square condition and the inherent condition. We then propose simple generic approximation algorithms for them and analyze their approximation ratios.

An Efficient Local Search for the Minimum Independent Dominating Set Problem

In the present paper, we propose an efficient local search for the minimum independent dominating set problem. We consider a local search that uses k-swap as the neighborhood operation. Given a feasible solution S, it is the operation of obtaining another feasible solution by dropping exactly k vertices from S and then by adding any number of vertices to it. We show that, when k=2, (resp., k=3 and a given solution is minimal with respect to 2-swap), we can find an improved solution in the neighborhood or conclude that no such solution exists in O(n\Delta) (resp., O(n\Delta^3)) time, where n denotes the number of vertices and \Delta denotes the maximum degree. We develop a metaheuristic algorithm that repeats the proposed local search and the plateau search iteratively. The algorithm is so effective that it updates the best-known upper bound for nine DIMACS graphs.

The Building Puzzle Is Still Hard Even in the Single Lined Version

The Building puzzle (a.k.a., the Skyscraper) is a Latin square completion-type puzzle like Sudoku, KenKen and Futoshiki. Recently, Iwamoto and Matsui showed the NP-completeness of the decision problem version of this puzzle, which asks whether a given instance has a solution or not. We provide a stronger result in the present paper; it is still NP-complete to decide whether we can complete a single line of the grid (i.e., a 1 × n or an n × 1 subgrid) without violating the rule.