Pith. sign in

REVIEW 3 major objections 4 minor 12 references

Continuous Game of Life: cell emergence and self-organization at the edge of growth

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A global resource constraint drives a continuous Game of Life to self-organize at the edge of growth.

desk verdict A genuinely new continuous GoL variant with a resource-feedback self-tuning mechanism; the qualitative edge-of-growth story holds up, but the quantitative 'within ~1%' claim rests on a fitted curve and needs a robustness pass. read the letter →

arxiv 2607.27402 v1 pith:76IMDJD4 submitted 2026-07-29 physics.bio-ph nlin.AOnlin.PS

Alexandre Guillet , Frank Jülicher This is my paper · ORCID
classification physics.bio-phnlin.AOnlin.PS MSC 37B1535B3692C15
keywords continuousGameofLifecellularautomatamorphogenesisself-organizationreaction-diffusionphasetransitionedgegrowthresourcefeedback
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper studies a minimal continuous version of Conway's Game of Life, in which a field relaxes toward a sigmoid-shaped target that depends on two Gaussian-blurred neighborhood fields. The model spontaneously produces cell-like patterns with a nucleus and shell that can divide, glide, oscillate, and die. The authors' central claim is that when growth consumes a finite resource, the system does not need its parameters tuned: the resource feedback itself retunes the dynamics until it sits at the boundary between a dilute phase of quiescent patterns and a dense space-filling phase, which they call the edge of growth. They argue that the reference parameters of the model lie within about one percent of this edge, and that near the edge the morphologies are the most diverse and life-like. A sympathetic reader would care because this offers a route from a simple, hand-built rule to self-organized life-like behavior without a carefully tuned control parameter.

What carries the argument

The central object is the target function Γ(M,N)=S′((N−Nc(M))/δNc(M)), the smooth analogue of the Game of Life survival rule, together with the equivalence rΓ(M,N;p) ∼ Γ(M,N;p/r), which identifies scaling the growth rate with scaling all target parameters. The resource feedback uses this equivalence: as cells grow, the abundance r(t)=Ra/R decreases, effectively retuning the parameters p/r(t) toward the dilute-to-dense transition. The edge itself is the boundary between the volume-limited dense phase and the resource-limited dilute phase; the paper locates it by fitting the equilibrium density to μ ≃ μ∞/√((ρ*/ρ)²+1), with ρ* the crossover resource density.

What would settle it

Run the resource-limited dynamics at increasing box sizes and runtimes, starting from the dense phase, and measure the equilibrium plateau of the abundance coefficient r(t). If the plateau position—and the fitted ρ* from Eq. (18)—moves systematically with domain size, integration time, or initial seed count, the claim that the system self-organizes to a well-defined edge of growth is falsified.

Watch

Extended reading notes

Core claim

The paper claims that a global conservation law for a finite resource—implemented as an abundance coefficient r(t)=Ra/R that scales the growth target or the morphogen fields—is sufficient to make the continuous Game of Life self-organize at the dilute-to-dense transition, the 'edge of growth.' Quantitatively, the hand-tuned reference parameters reside within about 1% of this edge: resource feedback type II retunes all six target parameters by the common factor r* = 0.987, and type III moves the first parameter by about 1%. At the edge, collective interactions can trigger divisions below the spontaneous growth threshold of an isolated cell, and a scan over the neighborhood scale ratio reveals

Load-bearing premise

The load-bearing premise is that the hand-fine-tuned reference parameters and the numerically fitted crossover density ρ* give a reliable location for the edge of growth; if that estimate shifts with box size, runtime, or initialization, the self-organization claim may be an artifact of the chosen numerical protocol.

Editorial extensions

If this is right

  • If correct, resource limitation replaces careful parameter tuning: starting the system in the dense phase is enough, because the feedback itself drives it to the edge of growth.
  • The reference parameters lie within about 1% of the edge, meaning the life-like phenomenology is not an isolated accident but sits on a boundary that organizes the phase diagram.
  • At the edge, collective interactions allow division below the single-cell growth threshold, so the dilute-to-dense transition is a collective effect rather than a single-particle property.
  • The model can be read as a coarse-grained reaction–diffusion system in which the target function specifies homeostatic morphogen ranges, and the Gaussian kernel shapes arise from a long cascade of fast auxiliary reactions.
  • In the large-volume limit at fixed total resource, the system's total mass saturates at μ∞R/ρ*, so the system becomes resource-limited and self-tuned rather than volume-limited.
  • Preliminary observations reported in the paper suggest that without resource feedback, evolution-like selection favors rapid spatial expansion, whereas resource limitation keeps the dynamics in a marginal regime where survival and reproduction depend on detailed pattern behavior.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: The edge-of-growth mechanism invites comparison with self-organized criticality, but the paper's own data show jumps, hysteresis, and finite-size sensitivity; a testable extension is to measure avalanche statistics and finite-size scaling to distinguish a first-order or coexistence-like transition from a truly critical one.
  • Editorial inference: The local mass-conserving version with finite resource diffusion is left as future work; one could test whether finite diffusion shifts the extracted edge location or alters cell motility, which would clarify how robust the global well-mixed result is.
  • Editorial inference: The paper's quantitative edge estimate relies on numerically fitted quantities (ρ*, μ∞) from finite-size simulations; a sharper test would derive the edge location from linear stability analysis of the homogeneous states rather than from fitted crossover densities.
  • Editorial inference: Since the (r, λ) scan samples only a slice of a likely high-dimensional edge-of-growth manifold, one could scan other target parameters to see whether the self-organization to the edge persists, and whether the reference parameters are special or one point on a larger organizing set.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper introduces a continuous-space, continuous-time variant of Conway's Game of Life, called cGoL, defined by an integro-differential equation with a bivariate survival rule Γ(M,N) and two Gaussian convolution kernels. The authors report a rich phenomenology of cell-like patterns that divide, glide, oscillate, and die, and relate these to homogeneous-state bifurcations, symmetry breaking, shape instabilities, and a dilute-to-dense collective transition. They also map the model onto a reaction–diffusion system with fast-relaxed morphogen-like auxiliary fields, and propose a resource-conservation feedback mechanism that dynamically retunes the growth threshold. The central claim is that, when resource limitation is introduced, the system self-organizes at the 'edge of growth' between dilute and dense phases, with the hand-tuned reference parameters of Eq. (7) lying within approximately 1% of this edge. The paper includes extensive numerical exploration of order parameters over a (r, λ) parameter plane and emphasizes the diversity of life-like morphologies found near the edge.

Significance. If the central edge-of-growth claim holds, the cGoL model would be a remarkably simple continuous system in which self-replicating, motile, localized patterns emerge and then self-tune to a dilute-to-dense boundary without external parameter tuning. The paper's strengths are its precise model definition, the detailed numerical phenomenology, the reaction–diffusion reinterpretation, and the explicit reporting of finite-size effects, seed dependence, and hysteresis. The data and code availability statement is a further positive feature. However, the quantitative centerpiece — that Eq. (7) is 'within ~1% of the edge of growth' — currently rests on a fitted resource-feedback saturation curve rather than on an independent measurement of the phase boundary. This makes the claim plausible but not yet established. The paper is exploratory and honest about its limitations, and the issues are addressable with additional analysis rather than being fatal.

major comments (3)
  1. [Sec. 5.3, Eq. (18)] The central quantitative claim that the reference parameters Eq. (7) lie 'within ~1% of the edge of growth' is underdetermined by the presented protocol. The value r* is not measured as an independent phase boundary; it is read off the resource-feedback plateau using the fitted relation μ ≃ μ∞/sqrt((ρ*/ρ)^2+1). In the resource-limited branch this relation gives r* = 1 − μ∞/ρ* by construction, with μ∞ and ρ* being fit parameters (Eq. 18) and with data points ν<4 excluded. The paper itself states in Sec. 6.4 that constant-r density cliffs overestimate r*, and Secs. 6.2, 6.4, and 6.5 report seed, time, and hysteresis dependence of edge estimates. Since Eq. (7) is rounded to two significant digits, the 1.3% separation between r*=0.987 and the reference r=1 is comparable to rounding and to the systematic uncertainties. To support the central claim, the edge should be located by an independent
  2. [Secs. 6.2, 6.4, 6.5] The edge estimates are finite-size- and protocol-dependent. The background-basin transition at λ=3 is stated to depend on time cutoff and volume (Sec. 6.2); the lower edge is described as a collective transition for which single-cell growth thresholds are 'useful but generally biased estimates' (Sec. 6.4); and Sec. 6.5 reports mass hysteresis and underestimates of r* when the resource is decreased. The self-organization claim is phrased in the large-volume limit, but the resource-feedback estimate is performed at fixed |Ω|=2^12. A finite-size study showing that the plateau value r* and the edge phenomenology are stable as |Ω| increases would substantially strengthen the central claim; alternatively, the claim should be explicitly limited to the simulated finite system.
  3. [Sec. 5.3, Eq. (18) and Sec. 6.4] The 'within ~1%' statement conflates the resource-feedback saturation value r* with the dilute-to-dense transition location. Eq. (18) is an empirical fit to equilibrium density data, not a derivation of a phase boundary. The paper's own Sec. 6.4 distinguishes the 'edge of growth' from the 'cliff' and notes that the cliff overestimates r*, while the true collective edge is influenced by collisions, seed number, run time, and hysteresis. Because the quantitative centerpiece depends on this distinction, the manuscript should provide a direct measurement of the edge at constant r (e.g., by varying r around 0.987 and measuring whether a dilute initial condition grows to the dense phase) rather than relying on the feedback plateau alone.
minor comments (4)
  1. [Eq. (18)] The fit 'excluding ν<4' is arbitrary and should be justified; reporting the fit range, residuals, and sensitivity to the exclusion threshold would help readers assess the reliability of μ∞ and ρ*.
  2. [Fig. 4] The color encoding of feedback types I, II, and III and the meaning of the grey asymptote lines should be stated explicitly in the caption or legend; the text refers to them but the figure description is incomplete.
  3. [Sec. 6.1, Eq. (20)] For a homogeneous field L=L*, the statement 'κ = dℓ → 0' should be written as 'κ ∼ dℓ → 0' to avoid the impression that the limit is taken before the bin size is sent to zero.
  4. [Secs. 3.3 and 6.4] The paper uses 'edge of growth' both for the resource-feedback saturation value and for the dilute-to-dense transition; these should be terminologically distinguished (e.g., 'feedback plateau' vs. 'phase boundary') to avoid circular reading.

Circularity Check

1 steps flagged · score 4.0 of 10

Quantitative edge-of-growth claim is fit-derived, but the qualitative mechanism is independently supported.

  1. fitted input called prediction [Sec. 5.3 (Eq. 18 and following), deployed in Secs. 6.4–6.5]
    "With feedback type II, all reference parameters are rescaled to p/r∗ by the common factor r∗ = 1 − µ∞/ρ(I,II)∗ ≃ 0.987, while feedback type III only modifies the first parameter to M∗ = (1 + µ∞/ρ(III)∗)Mc ≃ 1.010 Mc. Consequently, the hand-tuned reference parameters of Eq. (7) reside in the dense, volume-limited phase, within ∼ 1% of the edge of growth."

    Eq. (18) is introduced as an approximation with parameters µ∞ and ρ∗ 'used to fit the asymptotes' from equilibrium density simulations. The edge value r∗ is then not an independently measured dilute-to-dense phase boundary but the algebraic combination 1 − µ∞/ρ∗. In Sec. 6.4 the paper concedes that the constant-r cliff is a 'generally biased' estimate and that the 0.987 value 'relies on the self-tuning of r from the resource limitation feedback.' Thus the proximity claim reduces to the fitted parameters: change µ∞ or ρ∗, or the finite-volume protocol, and the '~1%' moves. It is a fit-derived quantity presented as the location of the edge, rather than an independent transition measurement.

full rationale

The paper's qualitative self-organization mechanism has independent numerical grounding: resource-limited simulations reach a plateau, constant-r scans show a dilute-to-dense transition in that region, and the morphology near the plateau is life-like. These observations are not merely definitional. However, the specific quantitative claim that Eq. (7) lies 'within ∼ 1% of the edge of growth' is computed from fitted quantities in Eq. (18): r∗ = 1 − µ∞/ρ∗, with µ∞ and ρ∗ fit to equilibrium density data. The paper itself flags finite-size effects, seed dependence, hysteresis, and biased cliff estimates (Secs. 6.2, 6.4, 6.5), which further undercut the independence of the 1% figure. No load-bearing self-citation chain is present; the references to prior work are external and the model derivation is self-contained. The circularity is therefore moderate and localized to the quantitative edge estimate, not to the core qualitative finding.

Assumptions & free parameters 5 free parameters · 6 assumptions · 3 invented entities

The central model behavior rests on a small hand-tuned parameter set, standard convolution/kernel identities, and numerical claims that resource feedback drives the system to a dilute-dense boundary. No external benchmark or formal proof anchors the edge-of-growth estimates; they are inferred from the same simulations used to define the transition.

free parameters (5)
  • Reference target parameters p = (Mc, δMc, N0, δN0, N1, δN1) = (0.50, 0.10, 0.23, 0.015, 0.35, 0.26)
    Hand-tuned and rounded so the elementary localized state divides several times, glides while oscillating, and disappears (Sec. 2.2).
  • Neighbourhood scale ratio λ = 3
    Fixed at the original GoL large-to-small neighborhood ratio; a geometric model parameter, later used as a second control in the phase scan.
  • Saturation density µ∞ = 0.131 ± 0.001
    Fitted asymptotic value of Eq. (18) from equilibrium simulations; used to compute r* and to argue the reference parameters lie within ~1% of the edge of growth.
  • Crossover resource density ρ* (feedback types I/II and III) = 10.5 ± 0.5 (I/II); 13.5 ± 0.5 (III)
    Fitted crossover parameters in Eq. (18); they set the self-tuned r* = 1 − µ∞/ρ*.
  • Average mass per cell m* and m∞ = m*(II)=6.10±0.05, m*(III)=6.05±0.05, m∞=5.725±0.005
    Fitted from cell-count and density data to relate ν, µ, R, and |Ω| (Sec. 5.3).
assumptions (6)
  • ad hoc to paper The survival rule Γ(M,N) = S′( (N−Nc(M))/δNc(M) ) with parameters Eq. (7) yields persistent cell-like patterns in the continuous-time limit.
    The parameters are fine-tuned and rounded to make the elementary state divide, glide, oscillate, and disappear (Sec. 2.2). This is the main empirical premise for the phenomenology.
  • domain assumption Numerical simulations on finite domains with fixed seeds and cutoffs represent the asymptotic attractors of the integro-differential equation.
    Phase scans use finite volume |Ω|, tmax=1000 or 5000, and 2–50 seeds; the authors report finite-size and initial-condition dependence in locating transitions (Secs. 6.2–6.5).
  • domain assumption Translation invariance and timescale separation allow auxiliary fields to be written as convolutions Ck = Φk * L.
    Used to map the cGoL to a reaction-diffusion system with fast-relaxing morphogens (Sec. 4.1); the timescale-separation limit is not proved for the simulated regime.
  • domain assumption The well-mixed limit Da→∞ makes the local conservation dynamics Eq. (17) equivalent to the global resource feedback Eq. (14).
    This bridges the global conservation law and the local mass-conserving interpretation; finite-diffusion dynamics is explicitly left to future work (Sec. 5.2).
  • standard math The long-cascade limit [1−σ²∇²]^{-k} ~ e^{k σ² ∇²} recovers Gaussian kernels (central limit theorem), with σ² ~ (4πk)^{-1} matching the unit-scale kernel.
    Formal Gaussian approximation used to interpret the model as a reaction cascade; valid for k≫1, not an exact identity for the simulated Gaussian kernels (Sec. 4.2).
  • domain assumption Cell number can be counted by thresholding M at Mc = 1/2 and counting connected components.
    Used to derive ν, m, and the resource scalings in Sec. 5.3; the threshold-to-cell correspondence is operational, not derived from first principles.
invented entities (3)
  • Morphogen fields M = Φ1 * L and N = Φ2 * L
    purpose: Auxiliary smoothed fields acting as continuous neighbor counts and interpreted as fast-relaxed morphogen concentrations controlling the slow field L.
    Introduced as mathematical auxiliary fields; the morphogen/homeostasis language is an interpretation (Sec. 4.1), not a claim about a physical chemical system.
  • Cascade of fast auxiliary species C_k (k≫1)
    purpose: Provides a reaction-diffusion mechanism whose long-cascade limit yields Gaussian kernels.
    Hypothetical linear reaction network (Eq. 10) used as an explanation for kernel shape; no independent experimental or observational evidence is provided.
  • Available-resource field ρa(x,t)
    purpose: Local mass-conserving counterpart to the global resource constraint; couples to L via Eq. (17).
    Introduced to embed the global feedback in a local conservation law; finite-diffusion effects are not studied (Sec. 5.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Continuous Game of Life: cell emergence and self-organization at the edge of growth." pith.science (2026). https://pith.science/paper/76IMDJD4

@misc{pith2026260727402,
  author       = {Pith},
  title        = {Pith review of: Continuous Game of Life: cell emergence and self-organization at the edge of growth},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/76IMDJD4}},
  note         = {Machine review of arXiv:2607.27402}
}
read the original abstract

Conway's Game of Life shows that simple rules can generate a rich diversity of emerging structures. This cellular automaton has been translated to continuous space by Rafler (2011) in a simulation called SmoothLife. The isotropic rule of this continuous Game of Life generates patterns whose beauty has attracted the attention of a growing community at the intersection of science and computer art. We study a minimal variant of this model, continuous in space and time, that generates cell-like patterns capable of self-replicating, gliding and disappearing. The phenomenology of these unit patterns is reported and related to homogeneous-state bifurcations, symmetry breaking, observed shape instabilities, finite-amplitude morphological changes, and a dilute-to-dense transition associated with cell proliferation. Its mapping onto a large reaction--diffusion system is interpreted in terms of homeostatic concentrations of morphogens, regulated by the nonlinear survival rule and generated through a cell-sourced cascade of auxiliary reactions. Introducing a global conservation law that limits resource availability causes the system to self-organize at this dilute-to-dense transition, which we call the edge of growth. A further exploration of parameter space reveals a variety of phases and the richness of life-like morphologies organized around this edge. Resemblance to biological processes such as division, motility, and death, together with a concise formulation and numerical implementation, makes the continuous Game of Life an appealing model system for investigating the emergence and self-organization of life-like patterns.

Figures

Figures reproduced from arXiv: 2607.27402 by the authors.

Figure 1
Figure 1. Primary and auxiliary fields: L, M = Φ1 ∗L and N = Φ2 ∗L (top, from left to right). Target function Γ (bottom left, same grey scale as for L, M, N). Special homogeneous states are marked in red: post-bifurcation unstable and stable (empty and full triangles) steady states, and pre-bifurcation non-steady optimum (empty circle). Spatial distribution of pairs (M, N), where Γ gets evaluated (bottom centre, pixel count).… view at source ↗
Figure 2
Figure 2. Cell dynamics at and near reference parameters Eq. (7), with colour coding of corresponding regions of the target function Γ (A) and the field L (B–E). Reference behaviours: dividing cell (B) or oscillating glider (C), depending on its symmetry axes (left). Time advances from left to right in steps of 10 (B) and 6 (C) time units. (D–E) Variants: (steady) isotropic egg (D) and (clockwise) turning glider (E) for N1 re… view at source ↗
Figure 3
Figure 3. Dynamic states of the cGoL model in the presence of a spatial variation of the target parameters p = p(x). For each panel, a subset of the parameters (specified on the left) is increased linearly from left to right around its reference value Eq. (7) (centre). The top panel corresponds to Eq. (8) with a position-dependent scaling r = r(x) (top axis). parameter Mc = 1 2 . The isotropic egg and turning glider variants … view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Equilibrium states as a function of resource R and volume |Ω|, in terms of the number of cells ν, the average density µ, and the mass per cell m. Lines encode different volumes (colours) and resource feedback types: target scaling (I), morphogen scaling (II), and kerne…
Figure 5
Figure 5. Figure 5: Scan of the density µ (order parameter), with constant resource abundance r as control parameter. Standard deviation around the mean (coloured area) over indepen￾dent simulations for a given volume |Ω|. Edges of growth and droplet stability (dotted and dashed lines). S…
Figure 6
Figure 6. Figure 6: Scan of order parameters (µ, κ) with control parameters (r, λ). The median over 20 trials is retained for |Ω| = 212, 5 λ-scaled seeds, tmax = 1000. Edge of growth (left, multiple estimates) and stability boundary for a single droplet (right) in the lower and upper back…
Figure 7
Figure 7. Figure 7: Resource-limited patterns, self-organized near the (lower) edge of growth, indi￾cated by blue markers in [PITH_FULL_IMAGE:figures/full_fig_p022_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 2 canonical work pages

  1. [1]

    gameofLife

    Akgün,H.,Yan,X.,Taşkıran,T.,Ibrahimi,M.,Lee,C.H.,&Jahangirov,S.(2026).Determin- istic scale-invariant dynamics in a logistic Game-of-Life model.Communications Physics,9(1),173. https://doi.org/10.1038/s42005-026-02568-w Alstrøm,P.,&Leão,J.(1994).Self-organizedcriticalityinthe“gameofLife”. Physical Re- viewE,49(4),R2507–R2508. https://doi.org/10/d2tx7j Bak...

    arXiv 2026

  2. [2]

    Chan, B. W.-C. (2019). Lenia: Biology of artificial life.Complex Systems, 28(3), 251–286. https://doi.org/10/ggf344

    2019

  3. [3]

    Chan, B. W.-C. (2020). Lenia and expanded universe.The 2020 Conference on Artificial Life,221–229. https://doi.org/10/gm3wrq CornusAmmonis.(2017,January).SmootherLife. https://www.shadertoy.com/view/XtVXzV

    2020

  4. [4]

    T., & Bongard, J

    Davis, Q. T., & Bongard, J. (2022). Step size is a consequential parameter in continuous cellular automata.The 2022 Conference on Artificial Life, 43.https://doi.org/10/ gs4dwp Evans,K.M.(2001).LargerthanLife:Digitalcreaturesinafamilyoftwo-dimensionalcellular automata.DiscreteMathematics&TheoreticalComputerScienceProceedings ,AA, 177–192.https://doi.org/1...

    2022

  5. [5]

    Gardner, M. (1970). Mathematical Games.Scientific American, 223(4), 120–123.https:// doi.org/10/cw53r6

    1970

  6. [6]

    Halatek, J., & Frey, E. (2018). Rethinking pattern formation in reaction–diffusion systems. NaturePhysics,14(5),507–514. https://doi.org/10/gcz7jc Hudcová, B., Dušek, F., Tuccio, M., & Hongler, C. (2026, January). Visualizing the structure ofLeniaparameterspace. https://doi.org/10.48550/arXiv.2601.01932 Hutton,T.,Munafo,R.,Trevorrow,A.,Rokicki,T.,&Wills,D...

    work page doi:10.48550/arxiv.2601.01932 2018

  7. [7]

    MacLennan, B. J. (1990, November).Continuous spatial automata(tech. rep. No. CS-90- 121).UniversityofTennessee,DepartmentofComputerScience.Knoxville,TN

    1990

  8. [8]

    Papadopoulos, V., Doat, G., Renard, A., & Hongler, C. (2024). Looking for complexity at phaseboundariesincontinuouscellularautomata. ProceedingsoftheGeneticand Evolutionary Computation Conference Companion, 179–182.https://doi.org/10/ hbdwtd

    2024

Show all 12 references
  1. [9]

    (2025, July)

    Papadopoulos, V., & Guichard, E. (2025, July). MaCE: General mass conserving dynamics forcellularautomata. https://doi.org/10.48550/arXiv.2507.12306 Pivato,M.(2007).RealLife:ThecontinuumlimitofLargerthanLifecellularautomata. The- oreticalComputerScience ,372(1),46–68. https://...

    2025 doi arXiv

  2. [10]

    Plantec, E., Hamon, G., Etcheverry, M., Oudeyer, P.-Y., Moulin-Frier, C., & Chan, B. W.-C. (2023). Flow-Lenia: Towards open-ended evolution in cellular automata through mass conservation and parameter localization.The 2023 Conference on Artificial Life,131. https://doi.org/10/g9w3x3

    2023

  3. [11]

    Game of Life

    Rafler, S. (2011, December). Generalization of Conway’s “Game of Life” to a continuous domain-SmoothLife. http://arxiv.org/abs/1111.1567 25 Rafler,S.(2012,June).SmoothLife. https://sourceforge.net/projects/smoothlife/files/ Reia,S.M.,&Kinouchi,O.(2014).Conway’sgameoflifeisanea...

    2011 arXiv

  4. [12]

    Yevenko, I., Kojima, H., & Nehaniv, C. L. (2025, August). Using dynamical systems theory to quantify complexity in asymptotic Lenia.https://doi.org/10.48550/arXiv.2508. 02935 26

    2025 doi

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.