TAG
#academic-research
0 篇评论 · 2 篇随笔
相关随笔
Can You Guard "One Key, One Door"? — Where Dungeon Generation Struggles, and Why a Puzzle About Stacking Colors Is NP-Hard
Two pieces today. "Evolutionary Wave Function Collapse," presented at IEEE CoG 2026 (Madrid, September 2), pairs evolutionary search with WFC, the go-to method for generating mazes and dungeons. It improves properties that emerge locally, like maze connectivity, but still struggles with constraints that span the whole level, like placing exactly one key and one door. The second piece, a paper by Linus Klocker at TU Wien in Austria, proves that Hexasort, a mobile puzzle where you stack and merge colored blocks, stays NP-hard even when restricted to a single color and tree-shaped boards. One paper generates; the other proves difficulty mathematically. Both answer the same question — where does a puzzle's difficulty actually come from — from different angles.
"Generate the Rules, Not the Levels" — RuleSweeper Has an AI Invent New Minesweeper Mechanics (IEEE CoG 2026)
One piece today: a look at RuleSweeper, presented at IEEE Conference on Games (CoG) 2026 (September 1-4, Madrid). Ryan Fleishman and colleagues at NYU had an LLM generate new rules for Minesweeper, not new boards, and ran a pipeline that tests each rule against a random agent, a symbolic solver, and an LLM-driven solver. Over 100 generations, 51 rule variants survived as genuinely playable games: mines that drift, mines that flash a warning first, clues that show relative rank instead of raw counts. It's a rare case of puzzle-generation research aiming at the rules themselves rather than just producing more levels.
