---
title: Non-Reciprocal Two-Species Vicsek Model
url: https://www.emergentmind.com/topics/non-reciprocal-two-species-vicsek-model
type: topic
---

# Non-Reciprocal Two-Species Vicsek Model

Searching arXiv for recent papers on non-reciprocal and two-species Vicsek models to ground the article.
The non-reciprocal two-species Vicsek model denotes a class of binary active-matter models in which two species of self-propelled particles undergo Vicsek-type orientational updates with asymmetric interspecies couplings, so that the influence of species \(A\) on species \(B\) differs from the influence of \(B\) on \(A\). In the strict sense used in recent work, non-reciprocity is implemented through unequal cross-couplings such as \(J_{AB}\neq J_{BA}\) or \(\chi^{12}\neq \chi^{21}\), while self-alignment is typically retained within each species. This class of models has been used to study chiral motion, demixing, large-scale structure formation, run-and-chase dynamics, and the breakdown of homogeneous flocking states [2604.18125]. Closely related binary Vicsek models with reciprocal antagonistic couplings or generalized multi-species extensions clarify which phenomena require explicit non-reciprocity and which already arise from heterogeneous but symmetric interactions [2211.10494].

## 1. Definition and scope of the model family

A canonical continuous-time non-reciprocal two-species Vicsek model in two dimensions assigns each particle \(i\) a position \(\mathbf r_i\), a heading \(\theta_i\), a species label \(s_i\in\{A,B\}\), constant speed \(v_0\), and metric interactions within a range \(R\). One representative formulation is
\[
\dot{\mathbf r}_i = v_0(\cos\theta_i,\sin\theta_i),
\]
\[
\dot{\theta}_i = \sum_{\alpha\in\{A,B\}} \frac{J_{s_i\alpha}}{N_i^\alpha} \sum_{j\in \mathcal N_i^\alpha}\sin(\theta_j-\theta_i) +\sqrt{2T}\,\xi_i(t),
\]
with \(J_{AA}=J_{BB}=1\) and \(J_{AB}\neq J_{BA}\) generating non-reciprocity [2604.06462]. A discrete-time metric version uses the standard noisy Vicsek update with a local weighted alignment field
\[
\bar{\hat e}_{i,s}^{t} = \frac{\sum_{s'} \sum_{j \in \mathcal{N}_i} J_{ss'} \,\hat{\mathbf e}_{j,s'}^{t}}{\left\|\sum_{s'} \sum_{j \in \mathcal{N}_i} J_{ss'} \,\hat{\mathbf e}_{j,s'}^{t}\right\|},
\]
followed by
\[
\theta_{i,s}^{t+\Delta t} = \arg\!\bigl(\bar{\hat e}_{i,s}^t\bigr)+\eta\,\xi_i^t,\qquad
\mathbf r_{i,s}^{t+\Delta t} = \mathbf r_{i,s}^{t}+v_s\,\hat{\mathbf e}_{i,s}^{t+\Delta t}\,\Delta t,
\]
with \(R_0=1\) in dimensionless units [2604.18125].

An alternative microscopic implementation uses discrete-time binary Vicsek mixtures with couplings \(\chi^{ab}\in[-1,1]\):
\[
\mathbf r_k(t+1)=\mathbf r_k(t)+v_0\,\hat{\mathbf u}[\theta_k(t+1)],
\]
\[
\theta_k(t+1)=\arg\!\left[\sum_{j\in \mathcal N_k}\chi^{ab}\,e^{i\theta_j(t)}+\eta\,n_k(t)e^{i\xi_k(t)}\right],
\]
where \(\chi^{11}=\chi^{22}=1\) and the interspecies couplings are decomposed into
\[
\bar\chi=\frac{\chi^{12}+\chi^{21}}{2},\qquad \Delta\chi=\frac{\chi^{12}-\chi^{21}}{2}.
\]
Here \(\Delta\chi\neq 0\) is the non-reciprocal part, and the regime of interest is weak nonreciprocity, \( |\Delta\chi|\le |\bar\chi| \) [2510.25377].

The defining distinction is that a genuine non-reciprocal two-species Vicsek model contains two physical species and an asymmetric \(A\to B\), \(B\to A\) coupling structure. This separates it from reciprocal two-species mixtures with \(J_{AB}=J_{BA}\), from single-species higher-order Vicsek models, and from models whose effective asymmetry arises only from local normalization or many-body effects rather than from an explicitly asymmetric coupling matrix [2512.19318].

## 2. Reciprocal, antagonistic, and non-reciprocal binary Vicsek models

The reciprocal baseline is the two-species Vicsek model (TSVM) of two unfriendly species, where particles of species \(A\) and \(B\) align with their own species and anti-align with the other. In that model the local interaction field is
\[
\bar{\boldsymbol{\sigma}_i^t} = \sum_{j\in\mathcal N_i} s_i s_j\,\boldsymbol{\sigma}_j^t,
\]
with \(s_i=\pm 1\), so same-species interactions are aligning and different-species interactions are anti-aligning [2211.10494]. The interaction is antagonistic but not non-reciprocal in the strict sense, because both species treat each other symmetrically through the same \(s_is_j\) factor.

A useful rewriting in the reciprocal TSVM introduces
\[
\boldsymbol{\alpha}_i \equiv s_i \boldsymbol{\sigma}_i,
\]
for which the update field becomes
\[
\bar{\boldsymbol{\alpha}_i^t} = \sum_{j\in\mathcal N_i}\boldsymbol{\alpha}_j^t.
\]
This rewriting clarifies the origin of parallel flocking (PF) and anti-parallel flocking (APF): PF is favored when the species segregate so that the “wrong” species is absent locally, while APF is favored when both species coexist with compatible \(\boldsymbol{\alpha}\)-ordering [2211.10494].

By contrast, strictly non-reciprocal models impose unequal cross-couplings. In one common antisymmetric case,
\[
J_{AA}=J_{BB}=J_{\rm self}>0,\qquad
J_{AB}=J_{\rm NR}>0,\qquad
J_{BA}=-J_{\rm NR}<0,
\]
so \(A\) aligns with \(B\) while \(B\) anti-aligns with \(A\), and the ratio
\[
\mu=\frac{J_{\rm NR}}{J_{\rm self}}
\]
measures the strength of non-reciprocal frustration [2604.18125]. A related weakly non-reciprocal formulation keeps both species mutually aligning or mutually anti-aligning on average, but with slightly different interspecies couplings, so that \(\bar\chi\) fixes the reciprocal part and \(\Delta\chi\) the asymmetry [2510.25377].

This terminological distinction matters because several nearby models are often conflated. The conformity-based higher-order Vicsek model derives effective pairwise and three-body terms from a single-species local-conformity rule,
\[
w_{jk}=a+\frac{b}{n_j}\sum_{l\sim j}\cos(\theta_l-\theta_k),
\]
leading to
\[
\dot{\theta}_j =\alpha r_j\sin(\psi_j-\theta_j) +\beta r_j q_j\sin(\phi_j-\psi_j-\theta_j) +\eta\xi_j(t),
\]
with \(\alpha=a-\tfrac{b}{2}\) and \(\beta=\tfrac{b}{2}\). Although its bidirectionally ordered state can be interpreted phenomenologically as two oppositely moving groups, it does not define two species or explicit asymmetric interspecies couplings [2512.19318]. Likewise, generalized reciprocal binary models with \(J_{AB}=J_{BA}\) can show rich phase coexistence, but they are not non-reciprocal at the coupling-matrix level [2503.17617].

## 3. Ordered states and collective phases

Reciprocal antagonistic binary Vicsek models exhibit a coexistence phenomenology that already extends the one-species Vicsek picture. In the TSVM, the flocking transition remains liquid-gas–like, with microphase separation into traveling dense bands moving through a gaseous background. Two band families occur: \(A\)-bands composed mainly of \(A\)-particles and \(B\)-bands composed mainly of \(B\)-particles. Within the coexistence region the system supports PF states, in which \(A\)- and \(B\)-bands move in the same direction, and APF states, in which they move in opposite directions [2211.10494].

These states are characterized by the order parameters
\[
\mathbf v_s(t)=\frac{1}{N}\sum_{i=1}^N \boldsymbol{\sigma}_i^t,\qquad
\mathbf v_a(t)=\frac{1}{N}\sum_{i=1}^N s_i^t\boldsymbol{\sigma}_i^t,
\]
with \(V_s=\langle|\mathbf v_s(t)|\rangle\) and \(V_a=\langle|\mathbf v_a(t)|\rangle\). In the thermodynamic limit PF has \(V_s>0\) and \(V_a=0\), whereas APF has \(V_s=0\) and \(V_a>0\); APF ordering is reported to be stronger than PF ordering [2211.10494].

Explicit non-reciprocity broadens the phase repertoire. In weakly non-reciprocal mixtures with \(\bar\chi>0\), increasing \(\Delta\chi\) destabilizes the homogeneous flock and produces a single traveling band mostly made of the more strongly aligning species, while the other species forms a more homogeneous liquid background. In the mutually anti-aligning case \(\bar\chi<0\), increasing \(\Delta\chi\) destabilizes homogeneous anti-flocking and drives laning, followed at larger asymmetry by polar clusters with chaotic dynamics and little macroscopic order [2510.25377].

In strongly non-reciprocal models, additional regimes appear. A continuous-time two-species Vicsek model with \(J_{AB}\neq J_{BA}\) shows a phase diagram containing homogeneous parallel flocking, homogeneous antiparallel flocking, longitudinal antiparallel lane, species-separated parallel-flocking band, and a chaotic phase. Near the antisymmetric line \(J_+=0\), the mean-field homogeneous chiral state is replaced by chaos at large scales [2604.06462]. In the discrete-time metric model with \(A\) aligning with \(B\) and \(B\) anti-aligning with \(A\), the observed regimes include chiral, aligned, anti-aligned, independently aligned, and weakly chiral states, with chirality occupying only a restricted window [2604.18125].

Closely related but formally distinct non-Vicsek or discrete-symmetry counterparts display analogous behavior. The non-reciprocal two-species active Ising model shows PF and one-species flocking at weak non-reciprocity, a disordered gas at intermediate frustration, and run-and-chase at strong non-reciprocity, with the latter interpreted as a traveling coupled-band state in which \(A\) chases \(B\) while \(B\) avoids \(A\) [2412.02501].

## 4. Hydrodynamic and kinetic descriptions

Several complementary coarse-grained descriptions have been developed. In weakly nonreciprocal mixtures, Boltzmann–Ginzburg–Landau coarse-graining introduces for each species a density field \(\rho^a(\mathbf r,t)\) and polarization field \(\mathbf p^a(\mathbf r,t)\), obeying
\[
\partial_t \rho^a + v_0 \nabla\cdot \mathbf p^a = 0,
\]
and
\[
\begin{aligned}
\partial_t p_i^a &+\sum_b \Big[ \psi_{ab}(\mathbf p^b\cdot\nabla)p_i^a +\lambda_{ab}(\nabla\cdot\mathbf p^b)p_i^a +\nu_{ab}\,\mathbf p^b\cdot\partial_i\mathbf p^a \Big] \\
&= \sum_b \Big[ \alpha_{ab}(\rho^1,\rho^2) -\sum_c \xi_{abc}(\mathbf p^c\cdot\mathbf p^c) \Big]p_i^a -\frac{v_0}{2}\partial_i \rho^a +\sum_b D_{ab}\nabla^2 p_i^a .
\end{aligned}
\]
The coefficient \(\alpha_{ab}\) depends on the densities, and this density dependence is central to the instability mechanism [2510.25377].

Linearization around homogeneous flocking or anti-flocking yields a \(6\times 6\) problem
\[
\partial_t \hat{\mathbf X}=\mathbb M(\mathbf q)\hat{\mathbf X},
\]
whose spectrum reveals two distinct instabilities: the usual Vicsek-type instability near onset with \(\mathrm{Re}(\Lambda)\sim q^2\) as \(q\to 0\), and a new non-reciprocal instability deeper in the ordered phase with \(\mathrm{Re}(\Lambda)\sim q\) as \(q\to 0\). The latter is the signature of a distinct long-wavelength mechanism [2510.25377].

In the \(q\to 0\) limit the dynamics reduces to three slow fields \((\delta\hat\rho^1,\delta\hat\rho^2,\bar\theta)\), with reduced matrix
\[
\partial_t \hat{\mathbf X}_{\rm h}=\mathbb M_{\rm h}(\mathbf q)\hat{\mathbf X}_{\rm h},
\]
and longitudinal eigenvalues
\[
\Lambda_\theta=iV_{\theta\theta}q_\parallel,\qquad
\Lambda_\pm=\frac{i q_\parallel}{2}\left(V_{11}+V_{22}\pm \sqrt{\mathcal R}\right),
\]
where
\[
\mathcal R = 4V_{12}V_{21} + (V_{11}-V_{22})^2.
\]
The instability criterion
\[
\mathcal R < 0 \quad\Longleftrightarrow\quad V_{12}V_{21}<-\frac{1}{4}(V_{11}-V_{22})^2
\]
requires the density cross-couplings to have opposite signs or sufficiently strong asymmetry, and is absent in reciprocal mixtures [2510.25377].

For the strongly non-reciprocal chiral problem, a Boltzmann kinetic theory expands the one-particle distributions \(f^\alpha(\mathbf r,\theta,t)\) in angular Fourier modes \(f_k^\alpha\), with density \(f_0^\alpha\) and complex polarization \(f_1^\alpha\). The hierarchy is truncated at \(k=4\), and stability of the homogeneous chiral limit cycle is analyzed via Floquet exponents \(\Lambda(\mathbf q)\). In all studied parameters, \(\Lambda(q)\) peaks at a finite wave number \(q_m\), so the instability of the homogeneous chiral state is finite-wavelength rather than long-wavelength [2604.06462].

A different but related kinetic framework has been derived for generalized multi-species Vicsek systems with additive sine couplings,
\[
\dot{\theta}_i = \sum_{j\in\mathcal N_i} J_{s(i),s(j)}\sin(\theta_j-\theta_i)+\sqrt{2D_r}\,\xi_i(t),
\]
using Fourier modes \(f_n^a(\mathbf r,t)\). In the reciprocal binary case, this theory predicts a finite-wavelength Turing-Hopf instability with \(k_c \approx 5.0\text{--}5.13\) and \(\lambda \approx 1.22\text{--}1.26\), matching a measured stripe wavelength \(\lambda = 1.222\pm 0.049\). The main binary treatment is reciprocal, but the general framework extends to directed cyclic couplings for \(m>2\) [2512.19393].

## 5. Instability mechanisms, demixing, and chirality

The principal mechanism identified in weakly non-reciprocal binary flocks is a feedback loop between local density fluctuations, local polarization, and species-dependent alignment strength. Because the local ordering coefficient \(\alpha_{ab}\) depends on density, a density fluctuation of one species modifies local order; if the couplings are non-reciprocal, the species respond differently, so one can polarize more strongly, advect differently, and amplify the original fluctuation. This mechanism is absent in a single-species flock and also absent in theories that ignore density fluctuations [2510.25377].

For mutually aligning species, this instability yields a single traveling band containing mostly one species, moving through a more homogeneous liquid of the other species. For mutually anti-aligning species, the same non-reciprocal mechanism destabilizes homogeneous anti-flocking and drives lanes, then polar clusters with large-scale chaotic dynamics. In both cases, species demixing appears without repulsive interactions [2510.25377].

Chirality constitutes a different non-reciprocal route. In the discrete-time metric model with antisymmetric interspecies couplings, the mean-field non-motile limit defines a phase-difference map
\[
\Delta^{t+1}=F(\Delta^t) = \arg\!\left(e^{i\Delta^t}+\mu\right) - \arg\!\left(1-\mu e^{i\Delta^t}\right),
\]
with fixed points
\[
\Delta^\star=0,\ \pi,\ \pm\frac{\pi}{2}.
\]
The branches \(\Delta^\star=\pm \frac{\pi}{2}\) are chiral, with mean-field angular velocity
\[
|\omega_{\rm MF}|=\arctan(\mu),
\]
and they are stable for any \(\mu>0\) in that non-motile mean-field picture [2604.18125].

In the motile finite-range model, however, chirality is not generic. It requires high density, very low motility, small system size relative to the interaction range, and sufficiently mixed local neighborhoods. The local composition indicator
\[
p_{\rm AA+BB}=\frac{n_{\rm AA}+n_{\rm BB}}{\sum n_{ss'}},\qquad p_{\rm AB+BA}=1-p_{\rm AA+BB}
\]
is close to \(0.5\) in the chiral regime, indicating strong local interspecies mixing [2604.18125]. This suggests that sustained mixing, rather than non-reciprocity alone, is necessary to synchronize the rotational frustration into a coherent chiral state.

A stronger statement is obtained in the continuous-time model: the homogeneous chiral state survives only below a characteristic length scale. The relevant intrinsic scale is the rotation radius
\[
r_{\rm rot}\sim \frac{v_0}{J_-},
\]
and the time-averaged chirality obeys a finite-size scaling form in the variable \(x=LJ_-/v_0\sim L/r_{\rm rot}\). For small \(x\) the chiral state is stable, while for large \(x\) the global chirality vanishes, indicating that the homogeneous chiral state is not asymptotically stable in large two-dimensional systems [2604.06462].

## 6. Finite-size effects, switching, and spatiotemporal chaos

Finite-size effects are central throughout the binary Vicsek literature. In the reciprocal TSVM, PF and APF can both exist in the low-density coexistence region and perform stochastic transitions from one to the other. The steady-state distribution \(P(v_s,v_a)\) becomes bimodal in this regime, and the dwell-time distribution has an exponential tail. At fixed aspect ratio \(L_x/L_y=8\), the PF dwell time scales as
\[
\tau_{\rm PF}\propto L_x^{z_1},\qquad z_1\simeq 1.7
\]
for small \(L_x\), then crosses over to
\[
\tau_{\rm PF}\sim L_x^{z_2},\qquad z_2\simeq 10,
\]
with an exponential scaling not ruled out. The crossover is controlled by the band width \(W\) relative to \(L_y\), and occurs roughly when \(W\approx L_y\) [2211.10494].

In non-reciprocal chiral flocks, finite-size control is even more stringent. Small systems can sustain long-lived coherent chirality, but larger systems show decay of the global chirality \(\chi(t)\) to zero, accompanied by species segregation and inhomogeneity. The decay-time distribution is inconsistent with simple Poisson nucleation of defects, which argues against a metastability picture based on independent droplet nucleation [2604.06462].

Beyond the finite-wavelength instability, the large-scale state is identified as extensive spatiotemporal chaos. The evidence includes a positive largest Lyapunov exponent, short-ranged velocity-velocity correlations, a broad energy spectrum, and an extensive number of unstable Floquet modes,
\[
N_+ \sim L^2.
\]
A master-slave synchronization protocol reveals a finite chaotic length scale: when the uncoupled core radius is smaller than this scale, the slave is entrained by the master, while for larger cores the slave develops independent chaotic dynamics. The crossover is found around
\[
r_{\rm core}\sim 10
\]
[2604.06462].

The complementary discrete-time metric study reaches a related conclusion from another angle. There, global chiral order weakens as \(L/R_0\) increases; for large systems the phase-difference correlation \(C_\phi(r)\) decays rapidly, indicating only local chirality. Population imbalance drives transitions from chiral states to porous PF or APF liquids, and motility imbalance induces asynchronous oscillations and, for sufficiently large mismatch,
\[
|v_{\mathrm A}-v_{\mathrm B}|\gtrsim 0.005,
\]
segregation into moving clusters of the faster species within a more dispersed background of slower particles [2604.18125].

## 7. Conceptual boundaries and related generalizations

A recurrent source of confusion is the tendency to classify any binary or bidirectional Vicsek-like ordering as non-reciprocal. Several recent models show why that identification is too broad. The local-conformity higher-order Vicsek model yields pairwise and three-body terms and supports a bidirectionally ordered phase with \(R<1\) and \(Q\simeq 1\), but it contains one species, symmetric metric interactions, no explicit two-population decomposition, and no asymmetric \(A\to B\), \(B\to A\) coupling matrix [2512.19318].

Likewise, generalized reciprocal two-species Vicsek models with
\[
J_{AA}=J_{BB},\qquad J_{AB}=J_{BA}
\]
already display parallel flocking, antiparallel flocking, flocking stripes, antiparallel flocking stripes, nematic stripes, and disordered hyperuniform phases. In that setting the couplings are reciprocal at the matrix level, even though local normalization by neighbor number can make the effective torques non-reciprocal in practice [2503.17617]. This suggests that some phenomena often associated with non-reciprocal active matter can also emerge from reciprocal competition between alignment and anti-alignment.

Other extensions broaden the field beyond the strict binary case. Metric-free topological mixtures with weak non-reciprocity, where species \(A\) aligns with both \(A\) and \(B\) while species \(B\) does not align with anyone, exhibit reentrant band–polar liquid–band behavior as a function of noise, and require higher angular harmonics for accurate coarse-grained stability analysis [2412.11871]. Non-reciprocal active Brownian mixtures with Vicsek-like torques and steric repulsion show asymmetric clustering in which single-species clusters chase more dilute accumulations of the other species; this is not a pure Vicsek point-particle model, but it demonstrates that broken action-reaction symmetry in orientational couplings alone can rotate the unstable density mode away from pure clustering or pure demixing [2403.19291].

At the multi-species level, cyclic directed couplings provide a fully non-reciprocal generalization. One framework uses a coupling graph
\[
J_{ss'}= \begin{cases}
-J, & s'=s,\\
J, & s'=(s+1)\pmod m,\\
0, & \text{otherwise},
\end{cases}
\]
while another introduces a constant inter-species phase shift \(\alpha\) and full \(S_Q\) Potts symmetry. These models generate chasing stripes, parity-dependent ordering, chirality, species separation, and coexistence phases, and clarify how the binary phenomenology extends to directed interaction networks [2512.19393].

Taken together, these results define the non-reciprocal two-species Vicsek model as a minimal but nontrivial active-matter setting in which asymmetry of interspecies alignment reorganizes flocking through density–polarization feedback, finite-wavelength or long-wavelength instabilities, and frustration-induced rotation. The strongest current conclusion is not that non-reciprocity generically produces a single universal ordered state, but that it opens several competing routes—demixing, chiral motion, bands, lanes, clusters, and chaos—whose selection depends sensitively on whether the asymmetry is weak or strong, reciprocal on average or antisymmetric, and compatible or incompatible with sustained local mixing [2510.25377].

Source: https://www.emergentmind.com/topics/non-reciprocal-two-species-vicsek-model