Kandel-Domany Approach in Cellular Automata

• The Kandel-Domany approach is a methodology using local, Markovian update rules in probabilistic cellular automata to analyze absorbing and active phase transitions.
• It employs mean-field closures, finite-size scaling, and cluster algorithms to derive rigorous bounds and understand critical behavior.
• Recent extensions include operator, tensor-network, and quantum reformulations, broadening its application across statistical mechanics and interdisciplinary models.

Searching arXiv for papers related to the “Kandel-Domany approach” and its uses across Domany-Kinzel, probabilistic cellular automata, cluster algorithms, and related statistical mechanics contexts. The Kandel-Domany approach is a label used for a cluster of methods centered on local stochastic update rules and their statistical-mechanical reformulations. In the Domany-Kinzel setting, it denotes the specification of local, Markovian transition probabilities $P(i|jk)$ for a $(1+1)$D probabilistic cellular automaton with an absorbing configuration, together with mean-field, scaling, and universality analyses of the resulting inactive-active transition (Boesl et al., 25 May 2026, Mendonça, 2015). Closely related developments recast the same logic in block-operator, zeta-function, tensor-network, quantum-cellular-automaton, FKG, and plaquette-cluster languages (Kiumi et al., 2022, Lesanovsky et al., 2018, Nigmatullin et al., 2021, Glazman et al., 2023, Capone et al., 21 Sep 2025). Recent arXiv usage also suggests that the expression is not fully standardized: some papers use it for a turbulence-statistics framework associated with Kandel, Lazarian, and Pogosyan, and some terminology is explicitly separate from the Domany-Kinzel line of work (Kandel et al., 2017, Hernández-Padilla et al., 2020, Lerner, 2012).

1. Classical probabilistic cellular automata formulation

In the Domany-Kinzel automaton, each site is either “active” $(1)$ or “empty” $(0)$, and the system evolves stochastically by local update probabilities $P(i|jk)$. The rule is specified by

$$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$

This local, Markovian update rule is identified as the Kandel-Domany approach in the stochastic-automaton setting. For $p_1=0$, the configuration with all sites empty is absorbing: once entered, the system stays there forever (Boesl et al., 25 May 2026).

The same framework underlies the interpretation of inactive-active transitions in related probabilistic cellular automata. In the noisy additive $p$-XOR PCA, the update rule is

$$x^{t+1}_i = \begin{cases} x^t_i \oplus x^t_{i+1} & \text{with probability } p,\ 0 & \text{with probability } (1-p), \end{cases}$$

and the model is equivalent to the Domany-Kinzel PCA on the line $p_2=0$ in DK parameter space. In this formulation, the inactive phase is the absorbing all-zero state, while the active phase has a stationary density of $(1+1)$0s, and the transition belongs to the directed percolation universality class (Mendonça, 2015).

A recurring feature is the relation between local update rules and absorbing-state criticality. In the Domany-Kinzel phase diagram, low $(1+1)$1 corresponds to the absorbing phase, high $(1+1)$2 to the active phase, and the transition is in the directed percolation class (Boesl et al., 25 May 2026).

2. Mean-field hierarchy, scaling, and rigorous bounds

A characteristic analytical component of the Kandel-Domany program is the combination of master equations, cluster mean-field closures, and finite-size scaling. For the $(1+1)$3-XOR PCA, the evolution of the configuration probability distribution is written as

$(1+1)$4

and the single-site mean-field closure

$(1+1)$5

leads to

$(1+1)$6

Its steady state predicts $(1+1)$7 or $(1+1)$8, with a critical point at $(1+1)$9. The pair approximation

$(1)$0

yields

$(1)$1

with $(1)$2. Monte Carlo and finite-size scaling give $(1)$3, with $(1)$4, $(1)$5, $(1)$6, and $(1)$7, in agreement within uncertainties with $(1)$8-dimensional directed percolation (Mendonça, 2015).

The scaling ansatz is written as

$(1)$9

which makes explicit how critical decay, finite size, and detuning from criticality are organized in the Kandel-Domany style of analysis (Mendonça, 2015).

More recent work turns these ideas into rigorous optimization problems. For synchronous and asynchronous nonequilibrium stochastic processes on infinite lattices, positivity of probability measures and invariance of invariant measures are used to formulate LP and SDP bounds. Applied to the Domany-Kinzel model, these bounds yield rigorous lower bounds on critical rates and two-sided bounds on the half-life of the infection density and the temporal correlation length in the subcritical phase. For the synchronous DK model with $(0)$0, the bound $(0)$1 is reported; for the asynchronous version, $(0)$2 is reported (Cho, 19 May 2025).

3. Operator, tensor-network, and quantum reformulations

One operator-theoretic reformulation writes the Domany-Kinzel dynamics in terms of local and global transfer operators. The local operator is

$(0)$3

and the global operator $(0)$4 is built recursively from tensor products. The evolution is $(0)$5, and the IPS-type zeta function is

$(0)$6

Its logarithm generates return-rate data through

$(0)$7

This makes block decomposition and trace recursion part of the Kandel-Domany toolkit (Kiumi et al., 2022).

A discrete-time quantum embedding is obtained from conditional gates

$(0)$8

with

$(0)$9

The reduced state at each time step is separable, yet for long times it becomes stationary and displays a continuous phase transition in the density of excited spins. Near the transition, density-density correlations become long-ranged, and quantum correlations quantified by the local quantum uncertainty also become long-ranged, despite separability. In some cases the local quantum uncertainty may be determined experimentally solely by measuring expectation values of classical observables (Lesanovsky et al., 2018).

A non-unitary quantum cellular automaton generalizes the Domany-Kinzel cellular automaton through a Lindblad master equation,

$P(i|jk)$0

with block-partitioned updates and neighbor-conditioned local three-site evolution. Numerical simulations using the iTEBD algorithm demonstrate an absorbing/percolating phase transition. Adding the Hamiltonian generates coherences, while the classical DKCA is recovered in the incoherent limit by tuning jump rates. A proposal for implementation with a Rydberg array is given, and it does not require local addressing of individual sites (Nigmatullin et al., 2021).

A further reformulation maps stochastic automata to isometric tensor network states. For a classical trajectory ensemble, the wavefunction is

$P(i|jk)$1

and the local Domany-Kinzel tensor is

$P(i|jk)$2

Probability conservation implies the isometry condition, so the network is an isoTNS. At the automaton’s critical point, the corresponding isoTNS hosts algebraic correlations in all spatial directions. For $P(i|jk)$3, the parent Hamiltonian has a degenerate ground-state manifold consisting of a product state and a second state that undergoes a transition from pairwise entanglement between distant regions, similarly to the $P(i|jk)$4 state, to a state with trivial entanglement (Boesl et al., 25 May 2026).

4. Correlation inequalities, planar duality, and plaquette cluster algorithms

Another important branch of the Kandel-Domany tradition is a percolation-based method built from Fortuin-Kasteleyn representations, cluster expansion, monotonicity via FKG, and planar duality. In recent two-dimensional work on the loop $P(i|jk)$5, six-vertex, and random-cluster models, this logic is extended through a novel joint FKG property for the triple $P(i|jk)$6. The Zhang-Sheffield non-coexistence theorem is then used to rule out simultaneous percolation of a configuration and its dual under planar translation-invariant FKG measures. The result is delocalisation up to the critical point, logarithmic fluctuation bounds in specific parameter regimes, and a proof of continuity of the random-cluster and Potts phase transition in two dimensions for $P(i|jk)$7 without parafermionic observables, Bethe Ansatz, or Russo-Seymour-Welsh theory (Glazman et al., 2023).

In Monte Carlo simulation, the Kandel-Domany algorithm generalizes cluster updates by replacing pairwise bond constructions with plaquette decompositions. For a frustrated long-range double-chain Ising model obtained from two antiferromagnetically coupled dissipative qubits after Trotter-Suzuky decomposition and integrating out the thermal bath, the action is decomposed as

$P(i|jk)$8

and plaquette graph weights $P(i|jk)$9 are chosen so that

$$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$0

This construction is designed to encode frustration locally. The conventional Swendsen-Wang approach becomes severely inefficient because frustration destroys the equality between spin correlations and cluster connectivity; by contrast, long-range plaquette decompositions are reported to be more efficient than local ones, especially for high levels of dissipation (Capone et al., 21 Sep 2025).

5. Geometric and evolutionary generalizations

The Domany-Kinzel framework has also been generalized away from flat, fixed-width fronts. In radial range expansions with mutation and selection, an off-lattice DK model is introduced to minimize persistent lattice artifacts. The generalized update rule uses local geometry and selective advantage: $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$1 At coarse-grained level, the local fraction $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$2 obeys

$$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$3

with $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$4. The characteristic crossover time is

$$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$5

after which portions of the colony become causally disconnected because of frontier inflation. Inflation suppresses continued coarsening, amplifies the effect of selective advantage relative to genetic drift, modifies the directed percolation transition, and introduces scaling functions analogous to a finite-size effect (Lavrentovich et al., 2012).

This generalization preserves the absorbing-state and directed-percolation logic of the Kandel-Domany approach while changing the causal structure. A plausible implication is that the method is less a single fixed model than a portable way of organizing stochastic local dynamics, coarse-grained field equations, and universality arguments across distinct geometries (Lavrentovich et al., 2012).

6. Terminological scope and common misconceptions

A common misconception is to treat the expression “Kandel-Domany approach” as if it named one uniquely defined formalism. The recent arXiv record suggests a more heterogeneous situation. In some papers the label denotes the Domany-Kinzel program of local stochastic rules, cluster closures, universality, FKG, and cluster algorithms; in others it is attached to unrelated or only terminologically adjacent frameworks.

Usage in the literature	Defining content	Representative arXiv
Domany-Kinzel local rule	Local, Markovian update probabilities $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$6 and absorbing-state dynamics	(Boesl et al., 25 May 2026)
Mean-field and scaling program	Cluster mean-field hierarchy, finite-size scaling, directed percolation identification	(Mendonça, 2015)
Percolation/FKG method	Fortuin-Kasteleyn representations, monotonicity, planar duality, non-coexistence	(Glazman et al., 2023)
Turbulence-statistics framework	$$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$7 decomposition, $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$8 cross-correlation, Alfvén/slow/fast modes, $$P(1|00)=p_1,\qquad P(1|01)=P(1|10)=p_2,\qquad P(1|11)=p_3 .$$9, $p_1=0$0	(Kandel et al., 2017)
Velocity-centroid anisotropy diagnostics	Mode decomposition and anisotropy of centroid structure functions in MHD turbulence	(Hernández-Padilla et al., 2020)
Dynamic limit-order-book terminology	Patient/impatient trader ratio, Markov law, Fokker-Planck evolution of $p_1=0$1	(Lerner, 2012)

The turbulence papers define the expression as the analytical statistical framework developed by Kandel, Lazarian, and Pogosyan for polarized dust and synchrotron emission, linking $p_1=0$2 power ratio, $p_1=0$3 cross-correlation, and anisotropy diagnostics to $p_1=0$4, $p_1=0$5, and density-magnetic-field correlation (Kandel et al., 2017, Hernández-Padilla et al., 2020). The market-microstructure paper associates the terminology with the dynamic Foucault, Kadan, and Kandel limit-order-book model, in which the ratio of patient and impatient traders drives waiting times, spreads, and volume-at-price distributions (Lerner, 2012). These usages are terminologically distinct from the Domany-Kinzel/statistical-mechanics lineage.

Taken together, the literature supports a narrow and a broad reading. In the narrow sense, the Kandel-Domany approach is the analysis of locally updated probabilistic cellular automata and their absorbing-state transitions. In the broader sense, this suggests a methodological style: encode local stochastic constraints, build an operator or cluster representation, and extract phase structure, correlations, or algorithmic updates from that representation.