Nivat's conjecture is a long-standing open problem in combinatorics that asks whether any two-dimensional pattern with sufficiently low complexity must be periodic in at least one direction. Here, complexity is measured by counting how many distinct shapes appear when you look at the pattern through a fixed window or template. The conjecture has been proven in some special cases, but finding or ruling out counterexamples for various window shapes remains an active area of research.
The authors construct a specific eight-cell window shape and a corresponding tiling of the integer grid that serves as a genuine counterexample to Nivat's conjecture for non-convex windows. Crucially, this counterexample is "non-degenerate," meaning it cannot be explained away by a simple number-theoretic trick involving the window being confined to a shifted sublattice, which settles a question posed by Kari and Moutot in 2023. On the positive side, the authors also prove that for windows whose size is the square of a prime number, every valid tiling does have a periodic one nearby, and they confirm that no window with fewer than eight cells can produce such a counterexample, with the single open case being windows of size six.