Cells That Remember the Curve: Non-Uniform Automata on Hyperbolic Lattices
16 May 2026, Yanjiang
A non-uniform cellular automaton on a hyperbolic lattice assigns unique rules to each cell, encoding the curvature and enabling exotic quantum symmetries unseen on flat grids.
We think of space as a flat, neutral backdrop—an empty stage upon which the laws of physics play out. But what if the stage itself were curved, twisted, expanding at an exponential rate, and its curvature could dictate what those laws even permit? That is the provocation at the heart of a new preprint (arXiv:2605.13379) from a team at Sun Yat-sen University. Xiang-You Huang, Jie-Yu Zhang, and Peng Ye have developed a framework to explore physics on hyperbolic lattices—surfaces with constant negative curvature where the number of neighboring cells balloons with every step—by teaching a cellular automaton to be non-uniform. Their results suggest that the geometry of a lattice can spawn exotic quantum phases and symmetries that have no parallel on a flat grid.
The challenge begins with a simple conceptual knot. A cellular automaton is a rule that updates each cell of a lattice based on the state of its neighbours—like a fabric whose pattern emerges from countless identical stitches. On a Euclidean square grid, the rule is the same everywhere: one size fits all. But roll that grid into a hyperbolic saddle, and the stitching falls apart. On a hyperbolic lattice, the number of nearest neighbours varies from cell to cell, and the connections multiply in a tree-like explosion. In such a landscape, the uniform rule that worked on a flat chessboard simply cannot apply. The result is not a crooked version of the flat fabric; it is a different textile altogether.
Enter the team’s non-uniform cellular automaton (NUCA). Their central insight is disarmingly elegant: instead of imposing a uniform rule on an unwelcoming geometry, they embed the hyperbolic lattice into a Euclidean square lattice through a “lattice-deforming procedure,” and then assign an update rule to each cell that carries the imprint of the original curvature. Think of this as a cartographer’s projection, where the distortions inherent in flattening a curved surface are woven into the rulebook itself—though in this case, the map deliberately sacrifices some neighbourhood relations to preserve the automaton’s computational structure. Each cell now obeys its own genetic code, one that remembers whether it sits at a branch point of the hyperbolic tree or deep in a quiet valley.
The payoff is immediate. With their NUCA in hand, the team set out to generate quantum states protected by subsystem symmetries—transformations that act only on a fraction of the system’s degrees of freedom, like local dress codes that leave the global constitution unchanged. On a flat Euclidean lattice, these symmetries tend to be tidy: they can be regular, fractal, or mixed. The NUCA on the hyperbolic lattice produces something far stranger. Guided by the treelike connectivity, the symmetries grow irregularly, sprouting arms at unpredictable rates. Some stretch at the same explosive pace as the lattice; others lag behind, creating asymmetry patterns that are neither fractal nor periodic. In technical language, the team realized subsystem symmetry-protected topological (SSPT) states with symmetries that have no counterpart on any Euclidean grid.
But wait—do these states represent a genuinely new kind of quantum order, or are they merely exotic projections of something already known? This question haunts the project from the start. The team addresses it by designing multi-point “strange correlators”—diagnostic tools that probe the hidden topology of the SSPT states. Their calculation shows that these correlators are non-trivial, a signature that the topological protection is robust and irreducible. An important tension sharpens the result: earlier work on subsystem symmetries in Euclidean systems had already produced fractal-like patterns, raising the suspicion that hyperbolic lattice symmetries might simply be Euclidean fractals in disguise. The Sun Yat-sen group’s systematic construction, however, offers a different answer. The irregular growth rates they observe are governed by the non-Abelian translation symmetry of the hyperbolic plane—an inherently non-commutative group that prevents any faithful Euclidean mapping. The symmetries are not decorated fractals; they are native citizens of curved space.
At this point, the tale takes a turn that could only happen in the language of cellular automata. The team generalised their NUCA to non-uniform Clifford quantum cellular automata—a form that accommodates the probabilistic logic of quantum bits. With it, they generated subsystem symmetries of the hyperbolic cluster state, a key resource for measurement-based quantum computation. This is not a mere extension; it transforms the NUCA from a tool for generating static phases into a blueprint for quantum information processing on hyperbolic surfaces. The Clifford NUCA stitches the cluster state’s entanglement together via non-uniform update rules, each step echoing the geometry of the underlying tree. The resulting symmetries are as intricate as the lattice itself, and they protect a form of quantum order that could one day be harnessed for error correction in a curved geometry.
Not content with pure quantum many-body landscapes, the researchers then aimed their NUCA at a classical problem that is deceptively simple to state and notoriously difficult to solve: directed percolation. Imagine a fluid that can only flow in a preferred direction, seeping through a porous medium. On a Euclidean lattice, the percolation threshold—the point at which the flow connects one side to the other—is a well-studied universal constant. On a hyperbolic lattice, the treelike structure accelerates the spread; pathways multiply so quickly that the threshold drops. By recasting the directed percolation process as a probabilistic NUCA, the team numerically mapped out the phase diagram and estimated critical probabilities. They found that the site percolation threshold sits around one-third—substantially lower than the familiar thresholds on flat square lattices. The result is a window into how geometry can make a network hyper-connected, a lesson that may resonate far beyond condensed matter.
And yet, the sceptic in the front row will already be raising a hand: can any of this be realised in a laboratory? The idea of a hyperbolic lattice is not a mathematical fantasy. Circuit quantum electrodynamics can engineer microwave resonators arranged in a hyperbolic tiling, and cold atoms in optical lattices can be trapped in effective curved metric potentials. This work, while theoretical, provides the rulebook that such experiments would need in order to program Hamiltonians with the desired symmetries. The NUCA algorithm is computationally light precisely because it trades uniformity for explicit geometric data—a bargain that experimentalists can exploit to simulate SSPT states without brute-force numerics.
What this work challenges is the quiet assumption that space, for all its grandeur, is merely a passive background. The geometry of the hyperbolic lattice is not a decoration painted onto the physics; it is a participant, capable of coaxing out symmetries that flat space forbids. In a sense, the team has held a mathematical prism up to nature and seen colours that our Euclidean eyes had never registered. The quantum phases they have catalogued are not just more of the same; they are creatures of a landscape where parallel lines diverge, where triangles get thinner as they grow, and where the very notion of “everywhere” starts to fray.
An important question, sharpened by earlier work on fractal subsystem symmetries, is whether the irregular symmetries reported here are truly intrinsic to the hyperbolic environment, or if they could be flattened onto a Euclidean lattice by sacrificing locality. The paper does not fully close that door, but it does erect a sturdy fence: the non-Abelian translation symmetry that underpins the NUCA construction is fundamentally incompatible with any Euclidean lattice unless one accepts non-local interactions. In other words, to get these symmetries on a flat chip, one would have to break the very locality that makes subsystem symmetries physically interesting. The tension between locality and geometry is not a flaw; it is the central message.
The road ahead is lit by the method alone. The NUCA algorithm is general: it extends readily to other hyperbolic tilings, such as the {6,6} honeycomb, and to higher dimensions. The team has essentially provided a programming language for geometry-driven quantum states. Whether the community will use it to discover new topological insulators, novel spin liquids, or error-correcting codes woven into curved spacetime depends on how comfortably physicists can learn to think in a curved, branching, non-uniform syntax.
Perhaps the deepest resonance of this work lies not in any particular SSPT state or percolation threshold, but in the revelation that the geometry of space can be a choreographer of quantum order. For decades, we have treated symmetry as something that emerges from the Hamiltonian, independent of the background. This preprint, in its methodical, constructive way, suggests that curvature can tip the balance, nudging nature toward phases that would otherwise never see the light of day. It is a reminder that the universe is not merely a container for physics; it is, in some subtle sense, an active ingredient. And if that is true, then the next generation of quantum simulators may need to abandon the flat drawing board and learn to build on spaces that breathe, branch, and remember their own shape.
Yanjiang is an online editor of LoomSci.com
References
- Xiang-You Huang et al., Universal Design and Physical Applications of Non-Uniform Cellular Automata on Translationally Invariant Lattices, arXiv:2605.13379