The Glider Equation for Asymptotic Lenia
Hiroki Kojima, Ivan Yevenko, Takashi Ikegami · 2025
Read The Glider Equation for Asymptotic Lenia by Hiroki Kojima, Ivan Yevenko, Takashi Ikegami (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-04167.txt.
Introduction
Hiroki Kojima, Ivan Yevenko, and Takashi Ikegami published a paper in 2025 deriving an equation that decides which patterns in a continuous Game of Life can glide. Lenia smooths Conway's classic cellular world into continuous space, and it breeds self-propelling blobs called gliders. In the variant studied here, the rules are partial differential equations, so calculus can attack what simulation only shows.
The authors derive the glider conditions exactly and name them the Glider Equation. Used as a loss function, the equation lets gradient descent hunt stable gliders on purpose, including faster variants, and a speed-free version widens the search to any velocity. Many finds eventually destabilize, yet the method surfaces formations blind search would miss. Links to neural field models close the paper.
Catalog record
| CREATOR | Hiroki Kojima, Ivan Yevenko, Takashi Ikegami |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-04167
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Glider Equation for Asymptotic Lenia Hiroki Kojima, Ivan Yevenko, Takashi Ikegami Abstract Lenia is a continuous extension of Conway’s Game of Life that exhibits rich pattern formations including self-propelling structures called gliders. In this paper, we focus on Asymptotic Lenia, a variant formulated as partial differential equations. By utilizing this mathematical formulation, we analytically derive the conditions for glider patterns, which we term the “Glider Equation.” We demonstrate that by using this equation as a loss function, gradient descent methods can successfully discover stable glider configurations. This approach enables the optimization of update rules to find novel gliders with specific properties, such as faster-moving variants. We also derive a velocity-free equation that characterizes gliders of any speed, expanding the search space for novel patterns. While many optimized patterns result in transient gliders that eventually destabilize, our approach effectively identifies diverse pattern formations that would be difficult to discover through traditional methods. Finally, we establish connections between Asymptotic Lenia and neural field models, highlighting mathematical relationships that bridge these systems and suggesting new directions for analyzing pattern formation in continuous dynamical systems. Keywords Lenia \cdot Continous Celullar Automata \cdot Neural Field \cdot Glider 1 Introduction Cellular Automata (CA) have been fundamental tools for studying emergent complexity from simple local rules. Conway’s Game of Life is a classic example that demonstrates how rich behaviors including self-replication and universal computation can emerge from elementary rules ( Gardner, 1970 ) . However, traditional CA are discrete in both space and time, limiting their connection to natural phenomena that often exhibit continuous dynamics. Lenia ( Chan, 2019 ; Chan, 2020 ) represents a significant advancement as a continuous generalization of cellular automata. Unlike the binary states and discrete updates of traditional CA, Lenia features continuous state values and smooth state transitions through the application of convolution kernels. This continuous framework allows for the emergence of diverse patterns with organic appearances, including self-propelling structures called “gliders” that resemble simple life forms. Recent work has further characterized the rich dynamics of Lenia and, for example, Yevenko, 2024 demonstrated the existence of fractal basins of attraction in Lenia’s pattern space, highlighting the system’s complex nonlinear behavior. While the original Lenia cannot be directly described by differential equations due to its clipping operation ( Davis and Bongard, 2022 ; Kojima and Ikegami, 2023 ) , variants have been proposed to bridge this gap. Asymptotic Lenia ( Kawaguchi et al., 2022 ) reformulates the system as partial differential equations (PDEs), making it amenable to analysis using the established mathematical frameworks of continuous dynamical systems. Using this formulation, Kojima and Ikegami, 2023 demonstrated that Asymptotic Lenia can be expressed as a reaction-diffusion system. Previous approaches to discovering glider patterns in Lenia-like systems have primarily relied on manual exploration, random search, or evolutionary algorithms ( Chan, 2020 ) . While these methods have successfully revealed interesting patterns, they lack the systematic approach that analytical formulations can provide. Recent work by Hamon et al., 2024 has explored the use of diversity search to discover sensorimotor agency in cellular automata, highlighting the richness of the pattern space that remains to be explored. The present work focuses on the PDE formulation of Asymptotic Lenia, leveraging this mathematical representation to analytically derive the condition for glider patterns, which we term the “Glider Equation.” This analytical approach provides new insights into the formation and properties of gliders in continuous cellular automata systems. Furthermore, we demonstrate how this equation can be used as the basis for an optimization framework to discover and design glider patterns with specific properties. The main contributions of this paper include the analytical derivation of the “Glider Equation” that characterizes self-propelling patterns in Asymptotic Lenia, along with a gradient-based optimization framework for systematically discovering diverse glider patterns. We further extend this framework to optimize update rules and discover patterns with specific properties, and develop a velocity-independent formulation of Glider Equation. Additionally, we identify and analyze connections between Asymptotic Lenia and neural field models, establishing links to established mathematical frameworks for studying pattern formation in continuous systems. The paper is organized as follows: In Section 2, we review the mathematical formulation of Asymptotic Lenia. Section 3 presents the analytical derivation of the Glider Equation. Section 4 introduces our optimization-based approach to pattern discovery. Section 5 develops the velocity-free formulation of the Glider Equation. Finally, we discuss the implications of our findings, including connections to neural field theory, and conclude with future directions. 2 Analytical Derivation of the Glider Equation 2.1 Mathematical Formulation of Asymptotic Lenia In two-dimensional space, the state evolution of Asymptotic Lenia is governed by: \frac{\partial}{\partial t}u(\mathbf{r},t)=T(K*u(\mathbf{r},t))-u(\mathbf{r},t) (1) where \mathbf{r}=(x,y) represents spatial coordinates. The kernel K typically has a radially symmetric form: K(\mathbf{r})=\frac{1}{Z}k(|\mathbf{r}|/R) (2) where R is the kernel radius, k is a kernel function, and Z is a normalization factor ensuring that \int K(\mathbf{r})d\mathbf{r}=1 . In practice, ring-shaped kernels are commonly used in Lenia systems, often constructed by stacking multiple concentric rings. In this work, we employ a three-ring kernel parameterization following the setup previously adopted in Asymptotic Lenia. The target function T is typically a “bell-shaped” function: T(x)=2\exp\left(-\frac{(x-\mu)^{2}}{2\sigma^{2}}\right)-1 (3) where \mu and \sigma control the center and width of the bell curve, respectively. In Lenia systems, the parameter \sigma is often set to small values, resulting in very sharp gaussian. 2.2 Gliders as Traveling Wave Solutions In cellular automata, gliders represent localized patterns that move across the grid while maintaining their structure. In discrete systems like Conway’s Game of Life, gliders typically undergo cyclic changes during movement. In contrast, in continuous cellular automata like Lenia, glider patterns that maintain their exact shape while translating across space are also observed. In this paper, we focused on this class of glider. This condition can be mathematically formulated as follows. If a pattern moves with a constant velocity vector \mathbf{v} without changing shape, then the value at \mathbf{r} at time t should be the same as the value at \mathbf{r}-\mathbf{v}t at time 0. This can be expressed as: u(\mathbf{r},t)=u(\mathbf{r}-\mathbf{v}t,0) (4) This is precisely the mathematical definition of a traveling wave solution. We note that the PDE formulation of Asymptotic Lenia enables us to establish this direct connection between gliders and traveling waves, opening the door to applying the rich theoretical framework of traveling wave analysis. While traveling wave theory has been extensively developed in various contexts, its application to Lenia-like systems represents a novel direction, as we will discuss further in relation to neural field models. By combining this traveling wave ansatz with the equations of Asymptotic Lenia, we can derive analytical conditions that any glider pattern must satisfy. Substituting the traveling wave ansatz into the Asymptotic Lenia equation requires computing the time derivative using the chain rule: \frac{\partial}{\partial t}u(\mathbf{r},t)=-\mathbf{v}\cdot\nabla u(\mathbf{r}-\mathbf{v}t,0) (5) Substituting back into the Asymptotic Lenia equation and defining \mathbf{r}^{\prime}=\mathbf{r}-\mathbf{v}t for simplicity: -\mathbf{v}\cdot\nabla u(\mathbf{r}^{\prime})=T(K*u(\mathbf{r}^{\prime}))-u(\mathbf{r}^{\prime}) (6) Rearranging and dropping the prime notation for clarity (as the equation must hold for all spatial positions), we obtain: u(\mathbf{r})-\mathbf{v}\cdot\nabla u(\mathbf{r})-T(K*u(\mathbf{r}))=0 (7) This equation, which we term the “Glider Equation,” represents the analytical condition for a pattern to maintain its shape while moving with velocity \mathbf{v} . The derivation marks a significant theoretical achievement: for the first time, we have an analytical characterization of glider patterns