Mass Conservation in Cellular Automata

• Mass Conservation in Cellular Automata is defined by local update rules that preserve site-variable sums, ensuring particle or mass invariance.
• The methodology employs discrete continuity equations and algebraic flow decompositions to rigorously enforce conservation laws at the microscale.
• These principles enable systematic enumeration of CA rules and practical simulations of transport phenomena in lattice gases, traffic flows, and hydrodynamics.

A cellular automaton (CA) is said to conserve mass (or particle number) if the sum of the relevant site variables—interpreted physically as mass, particle number, or density—remains strictly invariant under its local update rule. Mass conservation is a foundational property, making such automata paradigmatic models for lattice gases, classical transport, hydrodynamics, and other systems with explicit conservation laws. The study of mass-conserving cellular automata involves precise algebraic criteria, associated discrete continuity equations, structural classification of local rules, and rigorous connections between microscale invariance and emergent macroscopic transport phenomena.

1. Formal Criteria and Continuity Equations

Let $F: X \to X$ denote a CA on configuration space $X = A^{\mathbb{Z}^d}$, defined by a neighborhood $N$ and local rule $f : A^{|N|} \to A$. The CA possesses an additive conservation law if there exists a local density function $\rho_i : X \to \mathbb{Z}$ (typically a translate of a fixed local function $\rho$) such that for every finite region $\Lambda$,

$$\sum_{i \in \Lambda} \rho_i(F(x)) = \sum_{i \in \Lambda} \rho_i(x)$$

for all $x \in X$ (Metkar et al., 3 Jan 2026).

The necessary and sufficient local structure is encoded by the existence of a local current $j_{i+\alpha}$ (with finite support) enforcing a discrete continuity equation: $$\rho_i(F(x)) - \rho_i(x) = \sum_{\alpha} [j_{i-\alpha, i}(x) - j_{i, i+\alpha}(x)].$$ For $d=1$, this reduces to the classical difference form

$$m_i(t+1) - m_i(t) + j_{i+1/2}(t) - j_{i-1/2}(t) = 0.$$

The discrete divergence form is both necessary and sufficient for particle (mass) conservation (Metkar et al., 3 Jan 2026, Kari et al., 2013).

2. Algebraic Characterization and Structural Decomposition

In one dimension with states $A = \{0,1,\ldots,C\}$, all number-conserving rules admit a flow decomposition: $$\#(\varphi(a)) = \#(a_{r_1}) + f(a_{0:\ell-1}) - f(a_{1:\ell})$$ where $f$ is a flow function on length-$\ell$ neighborhoods, $\ell = r_1 + r_2$ [$2308.00060$]. The set of such flow functions forms a finite distributive lattice; minimal flows $m(a_1, a_2; k)$ serve as join-irreducible generators.

For general $d$-dimensional von Neumann CAs, any number-conserving local rule $f$ decomposes uniquely as $f = h + g$ where $h$ is a split function (fixed by its values on monomers, i.e., one-site excitations) and $g$ a perturbation (specified by basis vectors on dimers, i.e., two-site excitations with balanced antisymmetry) [$1901.05067$]:

• Split functions correspond to deterministic allocation of each site's mass among its neighbors.
• The space of perturbations is a vector space generated by linear combinations on dimers, subject to antisymmetry reflecting local conservation.

Necessary and sufficient constraints are provided (the "monomer-balance" and "dimer-balance" equations) allowing algorithmic enumeration of all number-conserving rules for given state set and dimension.

3. Mass Conservation on Diverse Lattices and Neighborhoods

On two-dimensional triangular and hexagonal lattices, the mass-conservation criterion reduces to a symmetric binary flow decomposition. For a local update $t(g,a,b,c)$ on a triangular lattice,

$$t(g,a,b,c) = g + \varphi(g, a) + \varphi(g, b) + \varphi(g, c)$$

with antisymmetric flow $\varphi(g, a) = -\varphi(a, g)$. Analogous formulas hold for permutation-symmetric hexagonal rules with a flow $\psi$ assigned to each center-neighbor pair [$0809.0355$].

In square-lattice binary CAs with the Moore neighborhood, mass conservation for a map $f : \{0,1\}^9 \to \{0,1\}$ is equivalent to a set of inclusion-exclusion constraints for every boundary-tail pattern (a total of 31 for the full Moore case), rendering the mass-invariant rule space tractable for systematic generation and compositional analysis (Wolnik et al., 9 Dec 2025).

4. Probabilistic and Stochastic Mass-Conserving Automata

Coarse-graining deterministic, mass-conserving CAs often induces stochasticity at the macro-level. In partitioned cellular automata (PCAs) built from local permutations, exact mass conservation is enforced microscopically. Coarse-graining by block projection yields emergent local rules that are typically convex combinations of mass-conserving permutations, forming an effective stochastic CA with mass conservation in expectation and outcome [$1905.10391$]. The appearance of random-walk–type behavior from underlying deterministic mass-conserving dynamics is intrinsic to the loss of microscopic information and underpins diffusive transport in lattice gases.

5. Connection to Invariant Measures and Statistical Physics

For surjective CAs, conservation of an additive Hamiltonian $H(\sigma) = \sum_i h(\sigma_i)$ is equivalent to invariance of the associated simplex of Gibbs equilibrium measures. The existence of a local discrete continuity equation (a cohomological equation for observables) guarantees that every shift-invariant Gibbs measure for $H$ is mapped onto itself by the CA, linking microscopic conservation to macroscopic stationarity of equilibrium states. In particular, Rule 184 preserves not only uniform Bernoulli but any product measure with fixed mean density (Kari et al., 2013).

This connection yields a unifying variational and cohomological characterization of conservation laws, both in terms of invariant measures and local current structures.

6. Algorithms, Enumeration, and Rule-Space Organization

Systematic construction and enumeration of mass-conserving CA rules rely on the explicit algebraic criteria described above:

• For $d=1$, form the lattice of permissible flows and construct rules via antichains of minimal flows [$2308.00060$].
• For $d>1$, enumerate all split functions and bounded perturbations, leveraging vector space structure to prune the rule space [$1901.05067$].
• For the Moore neighborhood, every nontrivial mass-conserving rule can be decomposed into an overall shift $\Omega$ and a set of conditional traffic-like (re-routing) regulations $\Lambda$, satisfying collision-prevention constraints (Wolnik et al., 9 Dec 2025).

Practical simulation leverages these structures, with memoization and antichain mutations facilitating efficient rule-space exploration and dynamics.

7. Extensions, Generalizations, and Applications

Continuous and hybrid models implement mass conservation in CA by local redistribution schemes. The MaCE framework achieves exact local and global mass conservation via a softmax-redistribution update rule, with a well-defined diffusion–advection PDE limit and high empirical soliton abundance (Papadopoulos et al., 16 Jul 2025). Flow Lenia extends this to continuous-state CA, embedding the parameter field into the mass-conserving advection step and enabling robust open-ended evolutionary dynamics in continuous media (Plantec et al., 2022).

Mass-conserving CAs underpin a wide range of physical models: lattice-gas automata, driven traffic flows, self-organized criticality, and kinetic transport theory. Algorithmic enumeration, compositional rule design, and transport diagnostics (via Green–Kubo formulas and scaling theory) collectively provide a rigorous basis for the modeling and analysis of discrete systems with intrinsic conservation laws (Metkar et al., 3 Jan 2026).

---

The mathematical rigor and explicit structural theorems governing mass conservation in CA unify its applications, allowing both systematic exploration of complex transport phenomena and principled design of discrete artificial systems constrained by conservation laws.