betaFeedback

WorksPhysics and Astronomy

Delay-induced anomalous density-polarization relation in chiral active matter

Abstract

Aligning active matter can spontaneously exhibit diverse forms of ordered collective motion, and higher local density is generally expected to enhance local polarization. Meanwhile, delays in agents' responses are widespread in real active systems, raising the question of how delayed alignment affects collective motion and, in particular, whether higher local density still strengthens polar order. Here, using a chiral active particle (CAP) model with delayed alignment, we uncover a counterintuitive anomalous density-polarization relation (DPR) within droplets: the local polarization first increases and then decreases with increasing local density. Analysis of the deviation-angle dynamics reveals a delay-induced supercritical discrete Hopf bifurcation. Noise-sustained oscillations before the bifurcation and self-sustained oscillations afterward broaden the deviation-angle distribution and reduce polarization. Increasing the delay shifts the bifurcation to lower densities, enhancing the anomalous DPR in agreement with simulations. Using the nonequilibrium landscape-flux theory, we further identify the nonequilibrium flux as the dynamical origin of the anomalous DPR. Our findings not only establish time delay as a control knob for collective behavior in chiral active matter, but also offer a useful analytical perspective on anomalous density-order relations in other nonequilibrium systems exhibiting collective order.

APSmanus.pdf

Active matter systems consist of agents that consume energy to generate directed motion [1]. They range from bacterial suspensions and bird flocks [2, 3] to artificial systems such as Janus particles and Quincke rollers [4, 5]. These systems exhibit diverse collective phenomena, including flocking, clustering, and active turbulence [6, 7, 8]. In aligning active matter, local interactions tend to align the orientations of neighboring agents. By bringing more neighbors within alignment range, higher density can favor polar order (e.g., in Quincke roller bands and mobbing jackdaw flocks [3, 9]). The Vicsek model [10] provides a minimal description of polar alignment. It has been extensively studied [11–15] and has spawned many variants [16–18]. Among these, Liebchen et al. Liebchen et al. [19] proposed a chiral variant of the Vicsek model, called the chiral active particle (CAP) model, which describes active matter systems with kinetic chirality commonly found in nature [20–23]. The introduction of chirality in the CAP model not only generates unique dynamic and thermodynamic behaviors [24–26] but also gives rise to rich patterns such as micro-flocks and macro-droplets [19, 27]. As we show below, local polarization increases monotonically with local density within droplets of the CAP model. We refer to this monotonic relation as the normal density–polarization relation (DPR).

In most Vicsek-like models (including the CAP model), an agent receives directional signals from its neighbors and changes its own direction simultaneously. In real-world systems, however, agents require time to respond due to the finite speeds of information transfer, processing, and actuation, which naturally introduce delays [28]. Remarkably, time delays have significant effects on active matter systems, such as inducing chirality [29] and forming collective patterns [30] in active Brownian particles with delayed attractions. Moreover, in flocking models [31] such as the Vicsek model [32, 33], time delays provide an effective means to switch between dynamical states, with short delays promoting order and long delays promoting disorder. These effects of delay on collective order motivate us to ask how delayed alignment reshapes the DPR within ordered droplets and what mechanisms underlie this change.

In this work, by incorporating delayed alignment into the CAP model, we develop a delayed CAP model and uncover a counterintuitive behavior within the formed droplets: local polarization first rises and then falls as local density increases, leaving the densest core less polarized than the periphery, contradicting the normal DPR in the delay-free CAP model. We term this counterintuitive behavior the anomalous density–polarization relation. Focusing on this behavior, we derive the dynamical equation for the deviation angle of particles inside droplets and analyze it via bifurcation theory [34]. We identify a supercritical discrete Hopf bifurcation in the deviation-angle dynamics, producing a periodic orbit with self-sustained oscillations above a critical density. Even below this density, noise-sustained oscillations occur around the stable fixed point. Both oscillations broaden the deviation-angle distribution and reduce polarization, accounting for the anomalous DPR. As the delay increases, the critical density decreases, shifting the bifurcation to lower densities, thereby making the anomalous DPR more pronounced, in agreement with the simulation results. Moreover, using the nonequilibrium landscape-flux theory [25, 35–39], we identify the nonequilibrium flux and its density derivative as the dynamical origin of the anomalous DPR and the signature of the bifurcation, respectively. In addition, we also explore the influence of the alignment strength and chirality strength on the anomalous DPR.

Model and Methods.– In the delayed CAP model, each chiral active particle experiences an alignment force after a delay τ\tau following the reception of neighbor signals and the computation of the alignment force. To be more precise, the dynamics of the ii-th particle is governed by

r˙i(t)=vni(t),\dot{\mathbf{r}}_i(t)=v\mathbf{n}_i(t),
θ˙i(t)=ω+KπR2∑j∈Nisin⁡(θj(t−τ)−θi(t−τ))+2ηζi(t)(1)\dot{\theta}_i(t)=\omega+\frac{K}{\pi R^2}\sum_{j\in N_i}\sin\left(\theta_j(t-\tau)-\theta_i(t-\tau)\right)+\sqrt{2\eta}\zeta_i(t) \tag*{(1)}

where ri\mathbf{r}_i, vv, θi\theta_i, and ni(t)=(cos⁡θi(t),sin⁡θi(t))\mathbf{n}_i(t)=(\cos\theta_i(t),\sin\theta_i(t)) are respectively the position vector, the self-propulsion speed, the angle, and the moving direction of the ii-th particle. The angular equation comprises chirality, alignment, and noise as its three terms. ω\omega is the chirality strength, representing the intrinsic rotation frequency of the particle. In the alignment term, KK and RR are the alignment strength and radius, respectively. NiN_i denotes the set of all particles in the neighborhood of particle ii (the region within RR), including particle ii itself. η\eta is the noise intensity, and ζi(t)\zeta_i(t) is the standard Gaussian white noise satisfying ⟨ζi(t)ζi′(t′)⟩=δii′δ(t−t′)\langle\zeta_i(t)\zeta_{i'}(t')\rangle=\delta_{ii'}\delta(t-t'). τ\tau denotes the delay time. For τ=0\tau=0, the delay effect disappears, and the model reduces to the original CAP model.

In our simulation, we use RR and R/vR/v as basic units for length and time, respectively. We discretize the equations (1) by using the Euler integration method and perform numerical simulations with N=32000N=32000 particles placed in a two-dimensional box of size L=80L=80 with periodic boundary conditions, corresponding to the global density ρ0=N/L2=5\rho_0=N/L^2=5. We set R=1R=1, v=1v=1, η=1.5\eta=1.5, and Δt=0.01\Delta t=0.01, where Δt\Delta t is the time step. We set ω=0.5\omega=0.5 and K=2K=2, if not otherwise stated. The initial construction of the delayed CAP system is random. All sampled data are collected after the system reaches steady state (10710^7 time steps).

Delay-induced anomalous DPR in ordered droplets.

To quantitatively describe the anomalous DPR, we introduce two parameters: the local density (ρL=n/A\rho_L=n/A) and the local polarization (ϕL=∣1n∑k∈Aeiθk∣\phi_L=\left|\frac{1}{n}\sum_{k\in A}e^{i\theta_k}\right|), where AA is the area of a local grid cell, and nn is the number of particles in AA. Accordingly, regions with ρL>ρ0\rho_L>\rho_0 and ρL<ρ0\rho_L<\rho_0 correspond to the ordered droplet and the disordered gas phase, respectively.

For the original and delayed CAP systems under our parameter setting, the ordered droplet phase can spontaneously emerge, as shown in Figure 1. In both systems, we observe that ρL\rho_L is higher in regions closer to the droplet core [Fig. 1(c,d)]. However, the spatial distribution of ϕL\phi_L differs between the two systems. In the undelayed system [Fig. 1(f)], regions with higher ρL\rho_L near the core also exhibit higher ϕL\phi_L. In contrast, in the delayed system [Fig. 1(e)], we observe a novel behavior: the region with the highest ρL\rho_L at the core actually shows a lower ϕL\phi_L than the peripheral regions.

Spatial distributions and density-polarization relations in ordered droplets

Figure 1. Anomalous density–polarization relation (DPR) induced by time delay. (a–f) Spatial distributions of the ordered droplet phase. (a,c,e) Snapshot, local density ρL\rho_L, and local polarization ϕL\phi_L, respectively, for the delayed CAP system (τ=0.05\tau=0.05). (b,d,f) Snapshot, ρL\rho_L, and ϕL\phi_L for the delay-free CAP system. (g) Mean local polarization ⟨ϕL⟩\langle\phi_L\rangle vs ρL\rho_L at various τ\tau. (h) ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} (orange), ⟨ϕL⟩(ρL,max⁡)\langle\phi_L\rangle(\rho_{L,\max}) (black), and Δϕ=⟨ϕL⟩max⁡−⟨ϕL⟩(ρL,max⁡)\Delta\phi=\langle\phi_L\rangle_{\max}-\langle\phi_L\rangle(\rho_{L,\max}) (red) as functions of τ\tau.

For clarity in describing the anomalous DPR, we plot the mean local polarization ⟨ϕL⟩\langle\phi_L\rangle against ρL\rho_L for various τ\tau in Figure 1(g). In the absence of delay (τ=0\tau=0), ⟨ϕL⟩\langle\phi_L\rangle increases monotonically with ρL\rho_L. This is consistent with the intuitive picture that a higher local density provides more neighbors, which strengthens the alignment force and thus enhances the local polarization. However, for nonzero τ\tau, the anomalous DPR is induced: ⟨ϕL⟩\langle\phi_L\rangle first increases and then decreases with increasing ρL\rho_L, exhibiting a non-monotonic dependence. As τ\tau increases, ⟨ϕL⟩\langle\phi_L\rangle decreases globally, and its peak shifts to lower ρL\rho_L, implying a more pronounced anomalous DPR.

To quantify the degree of the DPR anomaly, we define Δϕ=⟨ϕL⟩max⁡−⟨ϕL⟩(ρL,max⁡)\Delta\phi=\langle\phi_L\rangle_{\max}-\langle\phi_L\rangle(\rho_{L,\max}), which measures the polarization deficit of the densest region relative to the most polarized region within the droplet. Here, ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} is the maximum mean local polarization and ρL,max⁡\rho_{L,\max} is the maximal local density. Thus, Δϕ=0\Delta\phi=0 corresponds to a normal

monotonic DPR, whereas a larger Δϕ\Delta\phi indicates a more pronounced anomalous DPR. Figure 1(h) summarizes the dependence of ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} (orange), ⟨ϕL⟩(ρL,max⁡)\langle\phi_L\rangle(\rho_{L,\max}) (black), and Δϕ\Delta\phi (red) on τ\tau. The quantity ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} decreases linearly with increasing τ\tau, indicating that the delay reduces the overall polarization inside the droplet. Meanwhile, ⟨ϕL⟩(ρL,max⁡)\langle\phi_L\rangle(\rho_{L,\max}) decreases more rapidly than ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} does, but its decline slows down at larger τ\tau. Consequently, Δϕ\Delta\phi increases with τ\tau first, and tends to saturate at large delays, confirming that the delay enhances the anomalous DPR in general.

Dynamical bifurcation.— With the neighborhood quantities defined, the alignment term in the angular dynamics of Eq. (1), inspired by the Kuramoto model [40], takes the form: KρNi(t−τ)ϕNi(t−τ)sin⁡(αi(t−τ)−θi(t−τ))K\rho_N^i(t-\tau)\phi_N^i(t-\tau)\sin(\alpha_i(t-\tau)-\theta_i(t-\tau)), where ρNi:=∣Ni∣πR2\rho_N^i:=\frac{|N_i|}{\pi R^2}, ϕNi:=∣1∣Ni∣∑j∈Nieiθj∣\phi_N^i:=\left|\frac{1}{|N_i|}\sum_{j\in N_i}e^{\mathrm{i}\theta_j}\right|, and αi:=Arg⁡(∑j∈Nieiθj)\alpha_i:=\operatorname{Arg}\left(\sum_{j\in N_i}e^{\mathrm{i}\theta_j}\right) are the density, polarization, and average angle of CAPs in the ii-th particle’s neighborhood, respectively (see Supplemental Material (SM) for details). Inside the steady-state droplet, we define the deviation angle ψi(t):=θi(t)−αi(t)\psi_i(t):=\theta_i(t)-\alpha_i(t), and derive its evolution from Eq. (1) under the steady-state approximation (see SM for details):

ψ˙i(t)=−βisin⁡(ψi(t−τ))+2ηζi(t),(2)\dot{\psi}_i(t)=-\beta_i\sin(\psi_i(t-\tau))+\sqrt{2\eta}\zeta_i(t), \tag*{(2)}

where βi:=KρNiϕNi\beta_i:=K\rho_N^i\phi_N^i is the effective alignment strength of the neighborhood (ρNi\rho_N^i and ϕNi\phi_N^i are time-independent under our approximation explained in SM). Thus, the original multi-particle coupled Eq. (1) is reduced to an equivalent single-particle equation in the neighborhood, Eq. (2). Regardless of a particle’s position within the droplet, as long as it experiences the same ρN\rho_N and ϕN\phi_N, the evolution of its deviation angle ψ\psi follows Eq. (2).

To reveal the dynamical mechanism of anomalous DPR, we consider the deterministic part of Eq. (2). Discretizing it in time yields a single-variable delay-difference equation:

ψ(t+Δt)−ψ(t)=−βΔtsin⁡(ψ(t−τ)).(3)\psi(t+\Delta t)-\psi(t)=-\beta\Delta t\sin(\psi(t-\tau)). \tag*{(3)}

Through a standard bifurcation analysis [34], we characterize the dynamics of Eq. (3) (see SM for details). For τ>0\tau>0, at the critical point βc:=2Δtsin⁡(πΔt2(2τ+Δt))\beta_c:=\frac{2}{\Delta t}\sin\left(\frac{\pi\Delta t}{2(2\tau+\Delta t)}\right), the system undergoes a supercritical discrete Hopf bifurcation. In the limit Δt→0\Delta t\to0, βc→π2τ\beta_c\to\frac{\pi}{2\tau}. For 0<β<βc0<\beta<\beta_c, the fixed point ψ∗=0\psi^*=0 is locally asymptotically stable. For β>βc\beta>\beta_c, the fixed point loses stability and a periodic orbit emerges, corresponding to self-sustained oscillations. Linearizing the noisy version of Eq. (3) around ψ∗\psi^* predicts a polarization maximum at βs\beta_s (see SM for details). As Δt→0\Delta t\to0, βs/βc\beta_s/\beta_c approaches a constant (≈0.47\approx0.47). For βs<β<βc\beta_s<\beta<\beta_c, noise continuously excites the damped oscillatory mode that loses stability at βc\beta_c, sustaining quasicycles [41, 42]. As β\beta increases, weaker damping amplifies the response to noise, enhancing fluctuations in ψ\psi and reducing polarization (see Figure S2 in SM). In short, the dynamics of ψ\psi is determined by β\beta. Over the simulated

Figure S2. Deviation-angle power spectra S(Ω)S(\Omega) obtained from simulation data for βs<β<βc\beta_s < \beta< \beta_c. Parameters: τ=0.05\tau= 0.05, K=2K = 2, ω=0.5\omega= 0.5, and η=1.5\eta= 1.5.

density range, β\beta increases monotonically with ρN\rho_N (see Figure S3 in SM), giving a one-to-one correspondence. We thus define the optimal and critical densities ρs\rho_s and ρc\rho_c as the densities corresponding to βs\beta_s and βc\beta_c, respectively.

Equivalent alignment strength versus local density

Figure S3. The equivalent alignment strength β\beta versus the local density ρN\rho_N for τ=0.05\tau= 0.05.

To intuitively show the bifurcation, we plot the deterministic solution SdetS_{\mathrm{det}} of Eq. (3) and the steady-state probability distribution PP obtained from simulations, both in the (ψ(t−τ),ψ(t))(\psi(t-\tau),\psi(t)) phase plane for different ρN\rho_N. Take τ=0.05\tau=0.05 as an example. For ρs<ρN<ρc\rho_s<\rho_N<\rho_c [Figure 2(a,Figure 2b)], SdetS_{\mathrm{det}} remains a point and PP remains unimodal. As density increases, PP broadens, consistent with stronger quasicycles and reduced polarization. For ρN>ρc\rho_N>\rho_c [Figure 2(c)], SdetS_{\mathrm{det}} becomes a periodic orbit and PP develops a ring-shaped structure that expands with density, further reducing polarization. By contrast, for sufficiently large ρN\rho_N in the delay-free case, SdetS_{\mathrm{det}} remains a point and PP remains unimodal [Fig. 2(d)], which yields a narrow unimodal ψ\psi distribution and a large ϕN\phi_N, consistent with the normal DPR. Furthermore, the time autocorrelation of ψ\psi in the delayed CAP system displays three distinct behaviors: overdamped relaxation in the gas, and heavily and weakly damped oscillations in the droplet periphery and core, respectively (see SM for details).

Bifurcation, theoretical polarization curves, and scaled master curves for the anomalous DPR

Figure 2. Bifurcation, theoretical polarization curves, and scaled master curves for the anomalous DPR. (a–d) Probability distribution PP (color map) and deterministic solution SdetS_{\mathrm{det}} (red) in the (ψ(t−τ),ψ(t))(\psi(t-\tau),\psi(t)) phase plane, for neighborhood densities (a) ρN=9\rho_N=9, (b) ρN=17\rho_N=17, and (c) ρN=30\rho_N=30 with delay τ=0.05\tau=0.05, and (d) ρN=30\rho_N=30 without delay. (e) Theoretical polarization ⟨ϕN⟩max⁡\langle\phi_N\rangle_{\max} and fitted simulated polarization ⟨ϕN⟩\langle\phi_N\rangle versus ρN\rho_N for τ=0.05\tau=0.05. (f) Scaled polarization curves for different τ\tau in the scaled coordinates ρ~N\widetilde{\rho}_N and ϕ~N\widetilde{\phi}_N. Inset of (f): the maximum polarization ⟨ϕN⟩max⁡\langle\phi_N\rangle_{\max} (black dots) and the optimal density ρs\rho_s (orange dots) as functions of τ\tau; solid lines are fits. In (e, f), the gray and red vertical lines mark the optimal density ρs\rho_s and the critical density ρc\rho_c, respectively, and the red dashed line denotes the theoretical result.

Scaled curves.— For quantitative comparison with simulations, we compute the theoretical polarization ϕdist(β):=⟨cos⁡ψ⟩\phi_{\mathrm{dist}}(\beta) := \langle\cos\psi\rangle from the stationary distribution obtained numerically from Eq. (2). Combining ϕdist(β)\phi_{\mathrm{dist}}(\beta) with the corresponding density ρ:=β/[Kϕdist(β)]\rho:= \beta/[K\phi_{\mathrm{dist}}(\beta)] for each β\beta yields the theoretical DPR ϕdist\phi_{\mathrm{dist}}–ρ\rho. The theoretical curve ϕdist\phi_{\mathrm{dist}} agrees well with the simulation data ⟨ϕN⟩\langle\phi_N\rangle for τ=0.05\tau= 0.05 [Fig. 2(e)]. Moreover, ρs=βs/[Kϕdist(βs)]\rho_s = \beta_s/[K\phi_{\mathrm{dist}}(\beta_s)] coincides with the density at which ⟨ϕN⟩\langle\phi_N\rangle reaches its maximum, marking the onset of the polarization decline. Meanwhile, ρc=βc/[Kϕdist(βc)]\rho_c = \beta_c/[K\phi_{\mathrm{dist}}(\beta_c)] exactly coincides with the density at the most negative value of d⟨ϕN⟩/dρN\mathrm{d}\langle\phi_N\rangle/\mathrm{d}\rho_N (see Fig. S5 in SM), indicating the most rapid decrease of ϕN\phi_N with ρN\rho_N near ρc\rho_c. In addition, the ⟨ϕN⟩\langle\phi_N\rangle–ρN\rho_N curves for different τ\tau (see Fig. S6 in SM) are qualitatively consistent with the ⟨ϕL⟩\langle\phi_L\rangle–ρL\rho_L curves in Figure 1(g).

When plotted in scaled coordinates (ρ~N=(ρN−ρs)/ρs\tilde{\rho}_N = (\rho_N - \rho_s)/\rho_s, ϕ~N=⟨ϕN⟩/⟨ϕN⟩max⁡\tilde{\phi}_N = \langle\phi_N\rangle/\langle\phi_N\rangle_{\max}), the ⟨ϕN⟩\langle\phi_N\rangle–ρN\rho_N curves for different τ\tau merge onto a single master curve, as shown in Figure 2(f). The red dashed line represents the scaled theoretical curve ϕ~dist\tilde{\phi}_{\mathrm{dist}}–ρ~N\tilde{\rho}_N (see Figure S7 in SM), which also merges onto the same master curve. In scaled coordinates, the values of ρc\rho_c for different τ\tau nearly coincide, implying that ρs/ρc\rho_s/\rho_c is independent of τ\tau. The scaling parameters reflect the degree of the DPR anomaly (Δϕ\Delta\phi). As τ\tau increases, the vertical scale ⟨ϕN⟩max⁡\langle\phi_N\rangle_{\max} decreases linearly, whereas the horizontal scale ρs\rho_s decays as shifted τ−1\tau^{-1}, as shown in the inset of Figure 2(f). For small τ\tau, ρs\rho_s decreases much more rapidly than ⟨ϕN⟩max⁡\langle\phi_N\rangle_{\max}, resulting in a much stronger horizontal compression of the master curve than vertical. Consequently, ⟨ϕL⟩(ρL,max⁡)\langle\phi_L\rangle(\rho_{L,\max}) in Figure 1(h) drops faster than ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} with increasing τ\tau, causing a sharp increase in Δϕ\Delta\phi. For large τ\tau, the decay of ρs\rho_s slows down, resulting in a comparable vertical compression to horizontal compression. Hence, ⟨ϕL⟩(ρL,max⁡)\langle\phi_L\rangle(\rho_{L,\max}) and ⟨ϕL⟩max⁡\langle\phi_L\rangle_{\max} decrease at similar rates, thereby saturating the growth of Δϕ\Delta\phi.

Theoretical scaled curves of $\widetilde{\phi}_{\mathrm{dist}}$ versus $\widetilde{\rho}$ for nonzero delays

Figure S7. Theoretical scaled curves of ϕ~dist\widetilde{\phi}_{\mathrm{dist}} versus ρ~\widetilde{\rho} for nonzero delays. The gray and red lines mark ρ~=0\widetilde{\rho}=0 and ρ~=ρc/ρs−1\widetilde{\rho}=\rho_{\mathrm{c}}/\rho_{\mathrm{s}}-1, respectively, using ρs=7.6\rho_{\mathrm{s}}=7.6 and ρc=18.1\rho_{\mathrm{c}}=18.1 for τ=0.05\tau=0.05.

Dynamical mechanism of anomalous DPR.— To uncover the dynamical origin of the anomalous DPR, we apply the nonequilibrium landscape-flux theory [25, 35–39] in the (nτ+1)(n_\tau+ 1)-dimensional delay-embedded space spanned by ψ(t),ψ(t−Δt),…,ψ(t−nτΔt)\psi(t), \psi(t-\Delta t), \ldots, \psi(t-n_\tau\Delta t) at different ρN\rho_N, where nτ=τ/Δtn_\tau= \tau/\Delta t is the number of delay steps in the discretized system (see SM for details). In steady state, the probability density PP defines the nonequilibrium potential landscape U:=−ln⁡PU := -\ln P, which generates the gradient force Fgrad\mathbf{F}_{\mathrm{grad}}, tending to drive the system toward potential minima.

Figure 3. Landscape-flux analysis for τ=0.05\tau= 0.05. (a,b) 2D potential U2U_2 and flux J2\mathbf{J}_2 at ρN=9\rho_N = 9 and 30, respectively. (c) The averaged flux magnitude, JaJ_a; inset: Ja′=dJa/dρNJ'_a = \mathrm{d}J_a/\mathrm{d}\rho_N. (d) max Re⁡λ(Fcurl)\operatorname{Re}\lambda(\mathbf{F}_{\mathrm{curl}}) and max Re⁡λ(Fgrad)\operatorname{Re}\lambda(\mathbf{F}_{\mathrm{grad}}) for the curl (green) and gradient (orange) forces at different ρN\rho_N, respectively. Red and gray lines mark ρc\rho_c and ρs\rho_s, respectively.

potential minima. The nonequilibrium flux J:=Pv\mathbf{J} := P\mathbf{v} determines the curl force Fcurl\mathbf{F}_{\mathrm{curl}}, which drives circulation in this delay-embedded space. Here, v\mathbf{v} is the probability velocity field. For visualization, we perform dimensionality reduction and construct the two-dimensional potential U2=−ln⁡P2U_2 = -\ln P_2 and flux J2\mathbf{J}_2 in the (ψ(t−τ),ψ(t))(\psi(t-\tau),\psi(t)) plane (see SM for details). Below ρc\rho_c [xref]Fig. 3(a), J2\mathbf{J}_2 circulates around the stable fixed point in the single basin of U2U_2, tending to broaden the basin and counteracting the attraction toward the fixed point. Above ρc\rho_c [Fig. 3(b)], U2U_2 develops a ring-shaped valley that attracts the system toward the periodic orbit. J2\mathbf{J}_2 circulates along this valley, sustaining the oscillations.

The averaged flux magnitude, Ja:=⟨∣J∣⟩J_a := \langle|\mathbf{J}|\rangle, increases with ρN\rho_N [Fig. 3(c)], indicating stronger circulation that contributes to the polarization decline above ρs\rho_s. Moreover, the slope of the average flux, Ja′:=dJa/dρNJ'_a := \mathrm{d}J_a/\mathrm{d}\rho_N, reveals three distinct regimes [inset of Fig. 3(c)]. Below ρs\rho_s, the attractive basin dominates, while both Ja′J'_a and JaJ_a remain small, corresponding to the normal DPR regime. For ρs<ρN<ρc\rho_s < \rho_N < \rho_c, both Ja′J'_a and JaJ_a rise sharply with ρN\rho_N, so that the flux dominates, leading to a decrease in ϕN\phi_N and thus to the anomalous DPR regime. Ja′J'_a peaks near ρc\rho_c and then decreases slowly, indicating that JaJ_a grows most rapidly near the bifurcation and making Ja′J'_a a dynamical signature of it. Furthermore, we evaluate max⁡Re⁡λ\max\operatorname{Re}\lambda, the largest real part of the Jacobian eigenvalues for the gradient and curl forces (see SM for details). Near ρc\rho_c, max⁡Re⁡λ(Fgrad)<0\max\operatorname{Re}\lambda(\mathbf{F}_{\mathrm{grad}}) < 0 and max⁡Re⁡λ(Fcurl)>0\max\operatorname{Re}\lambda(\mathbf{F}_{\mathrm{curl}}) > 0 [Fig. 3(d)], revealing a competition between the stabilizing tendency of the potential landscape and the destabilizing tendency of the flux. Together, these results support the flux as the dynamical origin of the anomalous DPR.

Influence of KK and ω\omega on anomalous polarization.– In addition to τ\tau, we also examine the influence of KK and ω\omega on the anomalous DPR. Figure 4(a) shows Δϕ\Delta\phi for different values of KK and ω\omega. For fixed ω\omega, Δϕ\Delta\phi increases with KK, making the DPR more anomalous. The reason is that a larger KK raises β\beta, which facilitates the transition into the bifurcation regime. For fixed KK, Δϕ\Delta\phi first increases and then decreases with ω\omega, reaching a maximum Δϕmax⁡\Delta\phi_{\max} at ω∗≈0.6\omega^* \approx0.6 [Figure 4(a)]. This non-monotonic dependence cannot be explained by the equivalent single-particle equation Eq. (2) alone, because it neglects spatial coupling. In fact, it originates from the non-monotonic dependence of ρmax⁡\rho_{\max} on ω\omega, since ρmax⁡\rho_{\max} also peaks at ω∗\omega^* [Figure 4(b)]. When ω\omega is small, alignment interactions dominate the particle dynamics, allowing neighboring particles to rapidly phase-lock into a synchronously rotating macro-droplet. Due to the rotation-enhanced polarization effect [19], faster rotation enables the droplet to absorb randomly directed particles more effectively, thereby increasing its density. Consequently, for ω<ω∗\omega< \omega^*, ρmax⁡\rho_{\max} increases with ω\omega [Figure 4(b)]. However, as ω\omega increases further, the rotation timescale becomes comparable to or even shorter than the alignment timescale [19]. This generates significant phase differences between neighboring particles, frustrating phase-locking and driving the system into the micro-droplet phase [19]. These phase differences weaken polar alignment, reduce the droplet size, and consequently decrease the droplet density. Hence, for ω>ω∗\omega> \omega^*, ρmax⁡\rho_{\max} declines with ω\omega [Figure 4(b)]. Overall, ρmax⁡\rho_{\max} peaks at ω∗\omega^* for fixed KK, causing ⟨ϕL⟩(ρmax⁡)\langle\phi_L\rangle(\rho_{\max}) to reach its minimum, which in turn gives rise to a non-monotonic dependence of Δϕ\Delta\phi on ω\omega with a maximum at the same point.

Influence of $K$ and $\omega$ on the anomalous DPR

Figure 4. Influence of KK and ω\omega on the anomalous DPR. (a) Δϕ\Delta\phi in the KK–ω\omega parameter space. (b) ρmax⁡\rho_{\max} in the KK–ω\omega parameter space. Each data point is averaged over 20 ensembles. Parameters: τ=0.05\tau= 0.05, v=1v = 1, η=1.5\eta= 1.5.

Conclusion.– In summary, by incorporating time delay into the chiral active particle (CAP) model, we have uncovered a counterintuitive anomalous density-polarization relation (DPR) in ordered droplets: the densest core is less polarized than the lower-density periphery. As density increases beyond the optimal density ρs\rho_s, noise-sustained quasicycles around the stable fixed point broaden the deviation-angle distribution and reduce polarization. At the critical density ρc\rho_c, a supercritical discrete Hopf bifurcation gives rise to a periodic orbit, whose self-sustained oscillations further reduce polarization. Quantitatively, the density–polarization curve obtained from theory agrees well with those obtained from simulations. Furthermore, ρs\rho_s and ρc\rho_c decrease with increasing delay, bringing the bifurcation to lower densities, which makes the anomalous DPR more pronounced, in line with the simulation observation. More importantly, the nonequilibrium landscape-flux analysis identifies the nonequilibrium flux as the dynamical origin of the anomalous DPR and the density slope of the flux as a dynamical signature of the bifurcation. In addition, we explore the effects of alignment strength and chirality strength on the anomalous DPR. Our findings establish time delay as a mechanism for anomalous DPR and advance the theoretical understanding of collective behavior in chiral active matter. The resulting predictions are experimentally testable using programmable light-controlled robots [43, 44] and may guide the design of collective patterns. More broadly, our theoretical framework offers a route toward a unified understanding of density-modulated collective dynamics across diverse nonequilibrium systems, including polar active liquids [45], chemotactic bacteria [46], pedestrian crowds [47, 48], and confluent epithelia [49].

Dongrun Jian, Jie Su, and Jun Wang gratefully acknowledge funding from National Natural Science Foundation of China (Nos. 12404237 and 12234019), the Zhejiang Key Laboratory of Soft Matter Biomedical Materials (Nos. 2025ZY01036 and 2025E10072) and Wenzhou Institute of Biomaterials and Engineering (No. WIUCASQD2022012).

Supplemental Material: Delay-induced anomalous density-polarization relation in chiral active matter

Dongrun Jian; Jie Su; Jun Wang; Jin Wang

Derivation of the dynamical equation for the deviation angle

The parameter transformation

For a given particle ii, let NiN_i denote the set of all particles within the alignment radius RR of particle ii, including particle ii itself. We define the average orientation vector of the particles in NiN_i as:

ϕNieiαi=1∣Ni∣∑j∈Nieiθj,(4)\phi_N^i e^{i\alpha_i} = \frac{1}{|N_i|}\sum_{j\in N_i} e^{i\theta_j}, \tag*{(4)}

where θj\theta_j denotes the orientation angle of particle jj, ϕNi\phi_N^i is the neighborhood polarization, ∣Ni∣|N_i| is the number of particles in set NiN_i, and αi\alpha_i is the average angle over NiN_i. Multiplying both sides by ∣Ni∣πR2e−iθi\frac{|N_i|}{\pi R^2}e^{-i\theta_i}, taking the imaginary part, and replacing ∣Ni∣πR2\frac{|N_i|}{\pi R^2} by the neighborhood density ρNi\rho_N^i, we obtain the parameter transformation:

ρNiϕNisin⁡(αi−θi)=1πR2∑j∈Nisin⁡(θj−θi),(5)\rho_N^i\phi_N^i\sin(\alpha_i-\theta_i)=\frac{1}{\pi R^2}\sum_{j\in N_i}\sin(\theta_j-\theta_i), \tag*{(5)}

where θi−αi\theta_i-\alpha_i corresponds to the deviation angle ψi\psi_i used in the main text.

The steady-state approximation

Inside the steady-state droplet (ρNi>ρ0=5\rho_N^i>\rho_0=5), the neighborhood density ρNi(t)\rho_N^i(t) and polarization ϕNi(t)\phi_N^i(t) vary little over timescales of order τ\tau. We quantify their relative variations by

Rϕ:=⟨∣δϕNi(t)∣⟩⟨∣ϕNi(t)∣⟩,Rρ:=⟨∣δρNi(t)∣⟩⟨∣ρNi(t)∣⟩,(6)R_\phi:=\frac{\langle|\delta\phi_N^i(t)|\rangle}{\langle|\phi_N^i(t)|\rangle},\qquad R_\rho:=\frac{\langle|\delta\rho_N^i(t)|\rangle}{\langle|\rho_N^i(t)|\rangle}, \tag*{(6)}

where δϕNi(t):=ϕNi(t+τ)−ϕNi(t)\delta\phi_N^i(t):=\phi_N^i(t+\tau)-\phi_N^i(t), δρNi(t):=ρNi(t+τ)−ρNi(t)\delta\rho_N^i(t):=\rho_N^i(t+\tau)-\rho_N^i(t), and the averages are taken over time and particles. As shown in Figure S1, both RϕR_\phi and RρR_\rho are much smaller than 1 for ρNi>ρ0\rho_N^i>\rho_0, indicating that ϕNi\phi_N^i and ρNi\rho_N^i remain approximately constant over time intervals of order τ\tau. We therefore approximate them as time-independent, ρNi(t)=ρNi\rho_N^i(t)=\rho_N^i, ϕNi(t)=ϕNi\phi_N^i(t)=\phi_N^i.

Relative variations of the local polarization and local density, and ratio of angular velocities

Figure S1. (a) Relative variations of the local polarization RϕR_{\phi} and the local density RρR_{\rho} as functions of ρN\rho_N. (b) Ratio of ∣α˙−ω∣|\dot{\alpha}-\omega| to ∣ψ˙∣|\dot{\psi}|, RαR_{\alpha}, as a function of ρN\rho_N.

Defining the deviation angle ψi(t):=θi(t)−αi(t)\psi_i(t):=\theta_i(t)-\alpha_i(t), the angular equation Eq. (1) of the main text becomes

ψ˙i(t)=−βisin⁡(ψi(t−τ))+2ηζi(t)−[α˙i(t)−ω],(7)\dot{\psi}_i(t)=-\beta_i\sin(\psi_i(t-\tau))+\sqrt{2\eta}\zeta_i(t)-[\dot{\alpha}_i(t)-\omega], \tag*{(7)}

where βi:=KρNiϕNi\beta_i := K\rho_N^i\phi_N^i is the effective alignment strength of the neighborhood. Inside the droplet, α˙i(t)\dot{\alpha}_i(t) fluctuates weakly around ω\omega. To quantify the magnitude of this fluctuation relative to ψ˙i(t)\dot{\psi}_i(t), we define

Rα:=⟨∣α˙i(t)−ω∣⟩⟨∣ψ˙i(t)∣⟩,(8)R_{\alpha} := \frac{\left\langle\left|\dot{\alpha}_i(t)-\omega\right| \right\rangle}{\left\langle\left|\dot{\psi}_i(t)\right| \right\rangle}, \tag*{(8)}

where the averages are taken over time and particles. As shown in Fig. S1(b), Rα≪1R_{\alpha} \ll1 for ρN>ρ0\rho_N > \rho_0, indicating that ∣α˙i(t)−ω∣≪∣ψ˙i(t)∣|\dot{\alpha}_i(t)-\omega| \ll|\dot{\psi}_i(t)|. We may therefore neglect the term α˙i(t)−ω\dot{\alpha}_i(t)-\omega, yielding

ψ˙i(t)=−βisin⁡(ψi(t−τ))+2ηζi(t).(9)\dot{\psi}_i(t)=-\beta_i\sin\left(\psi_i(t-\tau)\right)+\sqrt{2\eta}\zeta_i(t). \tag*{(9)}

Bifurcation analysis of the deterministic delayed difference equation

To reveal the dynamical mechanism of anomalous DPR, we consider the deterministic part of (9). We define ψk:=ψ(kΔt)\psi_k := \psi(k\Delta t) and take τ=nτΔt\tau=n_{\tau}\Delta t, where the integer nτ≥0n_{\tau}\geq0 is the number of delay steps. Discretizing in time yields a single-variable delay-difference equation:

ψk+1−ψk=−βΔtsin⁡ψk−nτ.(10)\psi_{k+1}-\psi_k=-\beta\Delta t\sin\psi_{k-n_{\tau}}. \tag*{(10)}

The fixed points of (10) satisfy sin⁡ψ∗=0\sin\psi^*=0. We restrict the deviation angle ψ\psi to the range ψ∈(−π,π]\psi\in(-\pi,\pi], hence the fixed points are ψ∗=0\psi^*=0 or π\pi.

Introducing a small perturbation uk:=u(kΔt)u_k := u(k\Delta t), let ψk=ψ∗+uk\psi_k=\psi^*+u_k. The fixed point ψ∗=π\psi^*=\pi is unstable for any β>0\beta>0. Linearizing (10) around ψ∗=0\psi^*=0 yields the linear difference equation:

uk+1−uk=−βΔt uk−nτ.(11)u_{k+1}-u_k=-\beta\Delta t\,u_{k-n_{\tau}}. \tag*{(11)}

Substituting the ansatz uk=eλkΔtu_k=e^{\lambda k\Delta t}, where λ\lambda is the growth rate per unit time, yields the characteristic equation:

eλ(nτ+1)Δt−eλnτΔt+βΔt=0.(12)e^{\lambda(n_{\tau}+1)\Delta t}-e^{\lambda n_{\tau}\Delta t}+\beta\Delta t=0. \tag*{(12)}

We define the characteristic function by

L(λ):=eλΔt−1Δt+βe−λτ.(13)\mathcal{L}(\lambda):=\frac{e^{\lambda\Delta t}-1}{\Delta t}+\beta e^{-\lambda\tau}. \tag*{(13)}

(12) can then be written as L(λ)=0\mathcal{L}(\lambda)=0.

Taking the critical oscillatory eigenvalue λ=iΩ\lambda= \mathrm{i}\Omega, where Ω\Omega denotes the oscillation frequency, and substituting into Eq. (12), the phase and amplitude conditions give:

Ω(nτ+12)Δt=±π2(mod2π),βΔt=2∣sin⁡(ΩΔt2)∣(14)\Omega\left(n_{\tau}+\frac{1}{2}\right)\Delta t = \pm\frac{\pi}{2}\pmod{2\pi}, \qquad \beta\Delta t = 2\left\lvert\sin\left(\frac{\Omega\Delta t}{2}\right)\right\rvert \tag*{(14)}

where the ±\pm sign is matched with the sign of sin⁡(ΩΔt/2)\sin(\Omega\Delta t/2) (plus for sin⁡(ΩΔt/2)>0\sin(\Omega\Delta t/2)>0, minus otherwise), so that eiΩnτΔt(eiΩΔt−1)=−βΔte^{\mathrm{i}\Omega n_{\tau}\Delta t}(e^{\mathrm{i}\Omega\Delta t}-1)=-\beta\Delta t is a negative real number. Taking the lowest-frequency branch ΩcΔt=π2nτ+1\Omega_{\mathrm{c}}\Delta t=\frac{\pi}{2n_{\tau}+1} (with the plus sign), we obtain the bifurcation critical point:

βc=1Δt2sin⁡(π2(2nτ+1)).(15)\beta_{\mathrm{c}}=\frac{1}{\Delta t}2\sin\left(\frac{\pi}{2(2n_{\tau}+1)}\right). \tag*{(15)}

The following discrete Hopf analysis concerns nτ>0n_{\tau}>0. Let λ=λ(β)\lambda=\lambda(\beta). Differentiating the characteristic equation (S9) with respect to β\beta yields

dλdβ=−e−λτeλΔt−βτe−λτ=−1(nτ+1)eλ(nτ+1)Δt−nτeλnτΔt.(16)\frac{\mathrm{d}\lambda}{\mathrm{d}\beta} =-\frac{e^{-\lambda\tau}}{e^{\lambda\Delta t}-\beta\tau e^{-\lambda\tau}} =-\frac{1}{(n_{\tau}+1)e^{\lambda(n_{\tau}+1)\Delta t}-n_{\tau}e^{\lambda n_{\tau}\Delta t}}. \tag*{(16)}

Evaluating at the critical eigenvalue λ=iπ(2nτ+1)Δt\lambda=\mathrm{i}\frac{\pi}{(2n_{\tau}+1)\Delta t}, we obtain Re⁡(dλ/dβ)>0\operatorname{Re}(\mathrm{d}\lambda/\mathrm{d}\beta)>0, indicating that at the critical point βc\beta_{\mathrm{c}} the eigenvalue crosses the imaginary axis, confirming that this bifurcation is a discrete Hopf bifurcation.

Since the solution near the bifurcation point has its principal frequency near Ωc\Omega_{\mathrm{c}}, we project the solution onto the critical eigenmode eiΩckΔte^{\mathrm{i}\Omega_{\mathrm{c}}k\Delta t} (the complex amplitude representation):

ψk=zkeiΩckΔt+z‾ke−iΩckΔt+O(∣zk∣2).(17)\psi_k=z_k e^{\mathrm{i}\Omega_{\mathrm{c}}k\Delta t}+\overline{z}_k e^{-\mathrm{i}\Omega_{\mathrm{c}}k\Delta t}+O(\lvert z_k\rvert^2). \tag*{(17)}

Reducing the nonlinear delay-difference equation (S7) to the two-dimensional center manifold and retaining terms through cubic order, we obtain the normal form equation:

zk+1=zk+κ(β−βc)zk+νzk∣zk∣2+⋯ ,(18)z_{k+1}=z_k+\kappa(\beta-\beta_{\mathrm{c}})z_k+\nu z_k\lvert z_k\rvert^2+\cdots, \tag*{(18)}

where zkz_k is the complex amplitude at step kk, and κ\kappa and ν\nu denote the linear growth coefficient and the cubic saturation coefficient, respectively. Their real parts are explicitly given by Re⁡κ=(2nτ+1)βcΔt22d>0\operatorname{Re}\kappa=\frac{(2n_{\tau}+1)\beta_{\mathrm{c}}\Delta t^2}{2d}>0 and Re⁡ν=−(2nτ+1)βc2Δt24d<0\operatorname{Re}\nu=-\frac{(2n_{\tau}+1)\beta_{\mathrm{c}}^2\Delta t^2}{4d}<0, with d:=(2nτ+1)2βc2Δt24+cos⁡2ΩcΔt2d:=\frac{(2n_{\tau}+1)^2\beta_{\mathrm{c}}^2\Delta t^2}{4}+\cos^2\frac{\Omega_{\mathrm{c}}\Delta t}{2}. The first Lyapunov coefficient is l1:=Re⁡νl_1:=\operatorname{Re}\nu, whose negativity indicates that the system undergoes a supercritical discrete Hopf bifurcation at the critical point. Since the principal modal frequency is a rational fraction of π\pi, resonance occurs, forming a periodic orbit in delay-embedded space with a period of 2(2nτ+1)Δt2(2n_{\tau}+1)\Delta t (which can be regarded as a discrete version of a limit cycle).

In summary, the dynamical properties of the deterministic delayed difference equation are:

  1. For 0<β<βc0<\beta<\beta_{\mathrm{c}}, the fixed point ψ∗=0\psi^*=0 is locally asymptotically stable.

  1. At β=βc\beta=\beta_{\mathrm{c}}, the system undergoes a supercritical discrete Hopf bifurcation.

  1. For β>βc\beta>\beta_{\mathrm{c}}, the fixed point ψ∗=0\psi^*=0 loses stability and a period-2(2nτ+1)2(2n_{\tau}+1) orbit emerges, which is the “limit cycle” of the discrete system.

Steady-state variance and the maximum point of polarization

As in the main text, we define the theoretical polarization as ϕdist=⟨cos⁡ψ⟩\phi_{\mathrm{dist}}=\langle\cos\psi\rangle, where the average is taken over the stationary deviation-angle distribution at a given neighborhood density. Throughout this section, we consider fluctuations around the stable deterministic fixed point ψ∗=0\psi^* = 0 for 0<β<βc0 < \beta< \beta_c. Within the linear approximation, this distribution is a zero-mean Gaussian [1] with variance V:=Var⁡(ψ)V := \operatorname{Var}(\psi), yielding

ϕdist≃e−V/2.(19)\phi_{\mathrm{dist}} \simeq e^{-V/2}. \tag*{(19)}

Within this approximation, the polarization maximum corresponds to the variance minimum, and increasing variance implies decreasing polarization. We derive the stationary variance below and interpret its increase as β\beta approaches βc\beta_c from below.

Steady-state variance of the continuous system

Taking the continuous limit Δt→0\Delta t \to0 at fixed τ>0\tau> 0 in the discrete critical value Eq. (15), the bifurcation point becomes βc=π2τ\beta_c = \frac{\pi}{2\tau}. Linearizing the continuous dynamical equation Eq. (9) around ψ∗=0\psi^* = 0 yields:

u˙(t)=−βu(t−τ)+2ηζ(t).(20)\dot{u}(t) = -\beta u(t-\tau) + \sqrt{2\eta}\zeta(t). \tag*{(20)}

Its stationary variance V(β)V(\beta) can be derived from the Fokker–Planck equation [2, 1]:

V(β)=ηβ1+sin⁡(βτ)cos⁡(βτ),0<β<βc.(21)V(\beta) = \frac{\eta}{\beta}\frac{1+\sin(\beta\tau)}{\cos(\beta\tau)}, \qquad0 < \beta< \beta_c. \tag*{(21)}

Minimizing V(β)V(\beta) with respect to β\beta, i.e., setting V′′(β)>0V^{\prime\prime}(\beta) > 0 and V′(β)=0V'(\beta) = 0, gives cos⁡(βsτ)=βsτ\cos(\beta_s\tau) = \beta_s\tau, whose solution is the Dottie number βsτ=D≈0.739085\beta_s\tau= D \approx0.739085. Thus we obtain:

βsβc=2Dπ≈0.4705.(22)\frac{\beta_s}{\beta_c} = \frac{2D}{\pi} \approx0.4705. \tag*{(22)}

The ratio of the minimum point βs\beta_s to the bifurcation point βc\beta_c is a constant.

Steady-state variance of the discrete system

We now consider the discrete-time case. Writing uk:=u(kΔt)u_k := u(k\Delta t), the Euler–Maruyama discretization of Eq. (20) in the stable fixed-point regime (0<β<βc)(0 < \beta< \beta_c) reads:

uk+1−uk=−βΔt uk−nτ+2ηΔt ξk.(23)u_{k+1} - u_k = -\beta\Delta t\,u_{k-n_\tau} + \sqrt{2\eta\Delta t}\,\xi_k. \tag*{(23)}

Here ξk\xi_k are independent standard Gaussian variables, with ⟨ξk⟩=0\langle\xi_k\rangle= 0 and ⟨ξkξℓ⟩=δkℓ\langle\xi_k\xi_\ell\rangle= \delta_{k\ell}.

We now construct the Yule–Walker equations. For convenience, define γj:=Cov⁡(uk,uk−j)\gamma_j := \operatorname{Cov}(u_k,u_{k-j}), σ2:=Var⁡(2ηΔt ξk)=2ηΔt\sigma^2 := \operatorname{Var}(\sqrt{2\eta\Delta t}\,\xi_k) = 2\eta\Delta t, and β′:=βΔt\beta' := \beta\Delta t. Multiplying Eq. (23) by uk+1u_{k+1} and taking the expectation yields:

γ0−γ1=−β′γnτ+1+σ2.(24)\gamma_0 - \gamma_1 = -\beta'\gamma_{n_\tau+1} + \sigma^2. \tag*{(24)}

Since ξk\xi_k is independent of uk−ju_{k-j} for j≥0j \ge0, multiplying Eq. (23) by uk−ju_{k-j} and taking the expectation gives:

γj+1−γj=−β′γj−nτ,j=0,1,…,nτ.(25)\gamma_{j+1} - \gamma_j = -\beta'\gamma_{j-n_\tau}, \qquad j = 0,1,\ldots,n_\tau. \tag*{(25)}

Combined with the symmetry γj=γ−j\gamma_j = \gamma_{-j}, we obtain a closed linear system for γ0,γ1,…,γnτ+1\gamma_0,\gamma_1,\ldots,\gamma_{n_\tau+1}. Solving for γ0\gamma_0 yields the steady-state variance Vnτ(β′)V_{n_\tau}(\beta'). Vnτ(βΔt)V_{n_\tau}(\beta\Delta t) is a rational function of β\beta and possesses a minimum point βs\beta_s in the interval 0<β<βc0 < \beta< \beta_c. The steady-state variance minimum points βs\beta_s for different nτn_\tau are listed in Supplementary Table S1.

As nτn_\tau increases, βs/βc\beta_s/\beta_c approaches the constant value of the continuous limit (≈0.4705)(\approx0.4705). Moreover, in the continuous limit Δt→0\Delta t \to0 with τ=nτΔt\tau= n_\tau\Delta t fixed, the Yule–Walker solution reproduces the continuous variance of the previous subsection, Vnτ(βΔt)→V(β)V_{n_\tau}(\beta\Delta t) \to V(\beta), and correspondingly βs/βc→2D/π\beta_s/\beta_c \to2D/\pi. The steady-state variance functions Vnτ(β′)V_{n_\tau}(\beta') for discrete system are as follows:

For nτ=1n_\tau= 1:

V1(β′)=σ21+β′β′(2−β′−β′2).(26)V_1(\beta') = \sigma^2 \frac{1+\beta'}{\beta'(2-\beta'-\beta'^2)}. \tag*{(26)}
nτn_{\tau}βs\beta_sβc\beta_cβs/βc\beta_s/\beta_c
146.5571000.4656
228.95461.8030.4685
320.89244.5040.4694
416.31834.7300.4699
513.38028.4630.4701

Table S1. Steady-state variance minimum point βs\beta_s and its comparison with the critical value βc\beta_c for different delay steps nτn_{\tau}. All values are computed with Δt=0.01\Delta t = 0.01.

For nτ=2n_{\tau} = 2:

V2(β′)=σ21+β′−β′2β′(2−3β′−β′2+β′3).(27)V_2(\beta') = \sigma^2 \frac{1+\beta'-\beta'^{2}}{\beta'(2-3\beta'-\beta'^{2}+\beta'^{3})}. \tag*{(27)}

For nτ=3n_{\tau} = 3:

V3(β′)=σ21+2β′−β′2−β′3β′(2−3β′−4β′2+β′3+β′4).(28)V_3(\beta') = \sigma^2 \frac{1+2\beta'-\beta'^{2}-\beta'^{3}}{\beta'(2-3\beta'-4\beta'^{2}+\beta'^{3}+\beta'^{4})}. \tag*{(28)}

For nτ=4n_{\tau} = 4:

V4(β′)=σ21+2β′−3β′2−β′3+β′4β′(2−5β′−4β′2+5β′3+β′4−β′5).(29)V_4(\beta') = \sigma^2 \frac{1+2\beta'-3\beta'^{2}-\beta'^{3}+\beta'^{4}}{\beta'(2-5\beta'-4\beta'^{2}+5\beta'^{3}+\beta'^{4}-\beta'^{5})}. \tag*{(29)}

For nτ=5n_{\tau} = 5:

V5(β′)=σ21+3β′−3β′2−4β′3+β′4+β′5β′(2−5β′−9β′2+5β′3+6β′4−β′5−β′6).(30)V_5(\beta') = \sigma^2 \frac{1+3\beta'-3\beta'^{2}-4\beta'^{3}+\beta'^{4}+\beta'^{5}}{\beta'(2-5\beta'-9\beta'^{2}+5\beta'^{3}+6\beta'^{4}-\beta'^{5}-\beta'^{6})}. \tag*{(30)}

For nτ=0n_{\tau} = 0, the continuous-time variance V0=η/βV_0 = \eta/\beta decreases monotonically with β\beta. The finite minimum βs=1/Δt\beta_s = 1/\Delta t arises from Euler–Maruyama discretization and diverges as Δt→0\Delta t \to0. To avoid the resulting variance upturn, the delay-free comparison should satisfy βΔt<1\beta\Delta t < 1.

Noise-sustained oscillations below the bifurcation point

For βs<β<βc\beta_s < \beta< \beta_c, the deterministic fixed point remains stable, but noise continuously excites the damped oscillatory mode that becomes unstable at the bifurcation. These noise-sustained oscillations are termed quasicycles [3, 4]. To understand how these fluctuations reduce polarization, we examine their power spectrum S(Ω)S(\Omega). Fourier transforming Eq. (23) gives

S(Ω)=2η∣L(iΩ)∣2,(31)S(\Omega) = \frac{2\eta}{\lvert\mathcal{L}(i\Omega)\rvert^2}, \tag*{(31)}

where L\mathcal{L} is the characteristic function defined in Eq. (13), and ∣L(iΩ)∣−2\lvert\mathcal{L}(i\Omega)\rvert^{-2} gives the squared response to a unit harmonic perturbation at angular frequency Ω\Omega. The variance is the integrated spectral power: V=12π∫−π/Δtπ/ΔtS(Ω) dΩV = \frac{1}{2\pi}\int_{-\pi/\Delta t}^{\pi/\Delta t} S(\Omega)\,\mathrm{d}\Omega.

The oscillatory mode that loses stability at βc\beta_c has a characteristic root λm=−Γm+iΩm\lambda_m = -\Gamma_m + i\Omega_m, satisfying L(λm)=0\mathcal{L}(\lambda_m) = 0. Here, Γm>0\Gamma_m > 0 is its decay rate, and Ωm\Omega_m is its oscillation frequency, distinct from the spectral frequency variable Ω\Omega. As β\beta approaches βc\beta_c, Γm→0\Gamma_m \to0 and Ωm→Ωc\Omega_m \to\Omega_c. In the linear approximation, at fixed noise intensity, the resonant peaks have heights proportional to Γm−2\Gamma_m^{-2} and widths proportional to Γm\Gamma_m, so their integrated power grows as Γm−1\Gamma_m^{-1}. Thus, weaker damping amplifies the response to noise, increasing the variance of ψ\psi and reducing polarization. The predicted divergence at the threshold signals the breakdown of the linear approximation.

Figure S2 shows the power spectra S(Ω)S(\Omega) obtained from simulation data at three effective alignment strengths below βc\beta_c. As β\beta increases, the peak of S(Ω)S(\Omega) becomes more pronounced and the integrated spectral power increases. These observations support that noise sustains increasingly strong oscillatory fluctuations before the deterministic fixed point loses stability, broadening the deviation-angle distribution and lowering polarization.

Monotonicity of β(ρN)\beta(\rho_N)

Taking τ=0.05\tau= 0.05 as an example (Figure S3), the simulation data all fall within the monotonically increasing regime of β(ρN)\beta(\rho_N), ensuring a one-to-one correspondence between β\beta and ρN\rho_N.

Correlation in the delayed chiral active particle system

To capture the relaxation dynamics of ψ\psi, we compute its time autocorrelation function:

Cψ(t)=⟨cos⁡(ψi(t0+t)−ψi(t0))⟩i,t0,ρN∈W,(32)C_\psi(t) = \left\langle\cos\left(\psi_i(t_0+t)-\psi_i(t_0)\right)\right\rangle_{i,t_0,\rho_N\in W}, \tag*{(32)}

where the average is taken over time and over all particles ii whose neighborhood density ρN\rho_N lies within a prescribed window WW. We denote Cψg(t)C_\psi^{g}(t), Cψp(t)C_\psi^{p}(t), and Cψc(t)C_\psi^{c}(t) for the gas phase (W=(0,ρ0)W = (0,\rho_0)), the droplet periphery (W=(ρ0,ρc]W = (\rho_0,\rho_c]), and the droplet core (W=(ρc,ρmax⁡]W = (\rho_c,\rho_{\max}]), respectively. In the stationary regime, Cψ(t)C_\psi(t) approaches a long-time plateau:

C∞=⟨cos⁡ψ⟩2,(33)C_\infty= \langle\cos\psi\rangle^2, \tag*{(33)}

which corresponds to the theoretical average polarization ϕdist\phi_{\mathrm{dist}} of the particles in the window. To clearly compare different correlation behaviors, we define the normalized autocorrelation C~ψ(t)\widetilde{C}_\psi(t):

C~ψ(t)=Cψ(t)−C∞1−C∞.(34)\widetilde{C}_\psi(t) = \frac{C_\psi(t)-C_\infty}{1-C_\infty}. \tag*{(34)}

Take τ=0.05\tau= 0.05 as an example. The autocorrelation functions Cψ(t)C_\psi(t) for three different density windows exhibit different relaxation behaviors in Figure S4 (a). Cψc(t)C_\psi^c(t) shows a weakly damped oscillation that does not vanish within the observation time. This oscillation with period 0.220.22 is the temporal fingerprint of the period-22 orbit created by the bifurcation in the droplet core. Cψp(t)C_\psi^p(t) exhibits a heavily damped oscillation: its period is the same as that of Cψc(t)C_\psi^c(t), but the envelope shrinks quickly toward C∞pC_\infty^p, corresponding to the complex eigenvalues λ\lambda with Re⁡λ<0\operatorname{Re}\lambda< 0 of the linearized dynamical equation Eq. (11) (the stable fixed-point regime) in the droplet periphery. For the gas phase, Cψg(t)C_\psi^g(t) exhibits an overdamped relaxation, indicating that ψ\psi decorrelates from its history on a short timescale. Although Eq. (11) is only valid for the droplet phase (ρN>ρ0\rho_N > \rho_0), it can still qualitatively explain that the dominant eigenvalue λ\lambda becomes real and negative at low density, leading to the overdamped relaxation toward C∞gC_\infty^g. Moreover, as the density increases, the plateau C∞C_\infty first increases and then decreases, signaling the anomalous density–polarization relation reported in the main text.

Time autocorrelations of the deviation angle

Figure S4. Time autocorrelations of the deviation angle ψ\psi at τ=0.05\tau= 0.05. (a) Autocorrelation Cψ(t)C_\psi(t) (solid) with the plateau value C∞C_\infty (dashed). The superscripts gg, pp, and cc denote the gas, droplet-periphery, and droplet-core regimes, respectively. (b) Normalized autocorrelation C~ψc(t)\widetilde{C}_\psi^c(t) for different delays τ\tau.

In Figure S4 (b), we present the normalized autocorrelations in the droplet core C~ψc(t)\widetilde{C}_\psi^c(t) for different delays τ\tau. The oscillation period TcT_c can be written as 2(2nτ+1)Δt2(2n_\tau+1)\Delta t, precisely consistent with the period-2(2nτ+1)2(2n_\tau+1) orbit. Moreover, the oscillation envelope decays faster for larger τ\tau, indicating that the sustained oscillation is damped more strongly as the delay increases.

Derivative d⟨ϕN⟩/dρN\mathrm{d}\langle\phi_N\rangle/\mathrm{d}\rho_N versus the density $\rho_N`

In Figure S5, the theoretical polarization-maximum density ρs=7.6\rho_s = 7.6 lies close to the zero crossing of d⟨ϕN⟩/dρN\mathrm{d}\langle\phi_N\rangle/\mathrm{d}\rho_N, which identifies the polarization maximum in the simulation data. The bifurcation density ρc=18.1\rho_c = 18.1 agrees well with the minimum of d⟨ϕN⟩/dρN\mathrm{d}\langle\phi_N\rangle/\mathrm{d}\rho_N on the simulation curve, where d2⟨ϕN⟩/dρN2=0\mathrm{d}^2\langle\phi_N\rangle/\mathrm{d}\rho_N^2 = 0, indicating the most rapid decrease of ϕN\phi_N with ρN\rho_N near ρc\rho_c.

Figure S5. Derivative of the average polarization, d⟨ϕN⟩/dρN\mathrm{d}\langle\phi_N\rangle/\mathrm{d}\rho_N, as a function of the density ρN\rho_N for τ=0.05\tau= 0.05.

⟨ϕN⟩\langle\phi_N\rangle versus ρN\rho_N for different delays τ\tau

In Figure S6, unlike the delay-free case, the average neighborhood polarization ⟨ϕN⟩\langle\phi_N\rangle for τ>0\tau> 0 first increases and then decreases with increasing neighborhood density ρN\rho_N. As the delay τ\tau increases, ⟨ϕN⟩\langle\phi_N\rangle decreases globally, and its peak shifts to lower ρN\rho_N, implying a more pronounced anomalous density–polarization relation (DPR). The ⟨ϕN⟩\langle\phi_N\rangle–ρN\rho_N relation is qualitatively consistent with the ⟨ϕL⟩\langle\phi_L\rangle–ρL\rho_L curves in the main text.

Average neighborhood polarization versus neighborhood density for different delays

Figure S6. The average neighborhood polarization ⟨ϕN⟩\langle\phi_N\rangle versus the neighborhood density ρN\rho_N for different delays τ\tau

Scaling collapse of theoretical polarization curves

For the nonzero delays considered, the theoretical ϕdist(ρ)\phi_{\mathrm{dist}}(\rho) curves approximately collapse onto a single master curve when expressed in the scaled coordinates (ρ~=ρ−ρsρs,ϕ~dist=ϕdistϕmax⁡)(\widetilde{\rho}=\frac{\rho-\rho_{\mathrm{s}}}{\rho_{\mathrm{s}}},\widetilde{\phi}_{\mathrm{dist}}=\frac{\phi_{\mathrm{dist}}}{\phi_{\max}}), as shown in Figure S7.

Markov property of the delay-embedded dynamics

Markov process in the (nτ+1)(n_{\tau}+1)-dimensional delay-embedded space

Let ψk=ψ(kΔt)\psi_k=\psi(k\Delta t) and τ=nτΔt\tau=n_{\tau}\Delta t. The Euler–Maruyama discretization of Eq. (9) reads

ψk+1=ψk−βΔtsin⁡ψk−nτ+2ηΔt ξk,(35)\psi_{k+1}=\psi_k-\beta\Delta t\sin\psi_{k-n_{\tau}}+\sqrt{2\eta\Delta t}\,\xi_k, \tag*{(35)}

where ξk\xi_k are independent standard Gaussian variables, as in Eq. (23). We augment the state with the preceding nτn_{\tau} values of ψ\psi to form the (nτ+1)(n_{\tau}+1)-dimensional state vector

xk=(ψk,ψk−1,…,ψk−nτ)T.(36)\mathbf{x}_k=(\psi_k,\psi_{k-1},\ldots,\psi_{k-n_{\tau}})^{\mathsf{T}}. \tag*{(36)}

The dynamics of xk\mathbf{x}_k can be written as

xk+1=xk+Δt F(xk)+2ηΔt ξke0,F0(xk)=−βsin⁡ψk−nτ,Fr(xk)=ψk−r+1−ψk−rΔt,r=1,…,nτ,(37)\begin{aligned} \mathbf{x}_{k+1}&=\mathbf{x}_k+\Delta t\,\mathbf{F}(\mathbf{x}_k)+\sqrt{2\eta\Delta t}\,\xi_k\mathbf{e}_0,\\ F_0(\mathbf{x}_k)&=-\beta\sin\psi_{k-n_{\tau}},\\ F_r(\mathbf{x}_k)&=\frac{\psi_{k-r+1}-\psi_{k-r}}{\Delta t},\qquad r=1,\ldots,n_{\tau}, \tag*{(37)} \end{aligned}

where e0=(1,0,…,0)T\mathbf{e}_0=(1,0,\ldots,0)^{\mathsf{T}} and F(x)=(F0(x),…,Fnτ(x))T=E[xk+1−xk∣xk=x]/Δt\mathbf{F}(\mathbf{x})=(F_0(\mathbf{x}),\ldots,F_{n_{\tau}}(\mathbf{x}))^{\mathsf{T}}=\mathbb{E}[\mathbf{x}_{k+1}-\mathbf{x}_k\mid\mathbf{x}_k=\mathbf{x}]/\Delta t is the deterministic driving force.

The last nτn_{\tau} components of xk+1\mathbf{x}_{k+1} are deterministic shifts of the first nτn_{\tau} components of xk\mathbf{x}_k. The first component depends on xk\mathbf{x}_k and the new noise ξk\xi_k, which is independent of the current and all previous states. Thus, the discretized reduced equation Eq. (35) defines a Markov process in the (nτ+1)(n_{\tau}+1)-dimensional delay-embedded space.

Numerical test of the Markov property in the delay-embedded space

We set τ=0.05\tau= 0.05 and Δt=0.01\Delta t = 0.01 and test the Markov property in delay-embedded spaces of different dimensions, focusing on the six-dimensional case. We quantify history dependence in the embedded trajectories using conditional mutual information (CMI):

Im:=I(xk−1;xk+1∣xk)=I(ψk−m;ψk+1∣xk)=∫p(ψk−m,ψk+1,xk)ln⁡p(ψk−m,ψk+1∣xk)p(ψk−m∣xk)p(ψk+1∣xk) dψk−mdψk+1dxk,(38)\begin{aligned} I_m &:= I(\mathbf{x}_{k-1};\mathbf{x}_{k+1}\mid\mathbf{x}_k) \\ &= I(\psi_{k-m};\psi_{k+1}\mid\mathbf{x}_k) \\ &= \int p(\psi_{k-m},\psi_{k+1},\mathbf{x}_k)\ln\frac{p(\psi_{k-m},\psi_{k+1}\mid\mathbf{x}_k)}{p(\psi_{k-m}\mid\mathbf{x}_k)p(\psi_{k+1}\mid\mathbf{x}_k)}\,\mathrm{d}\psi_{k-m}\mathrm{d}\psi_{k+1}\mathrm{d}\mathbf{x}_k, \tag*{(38)} \end{aligned}

where xk=(ψk,…,ψk−m+1)T\mathbf{x}_k=(\psi_k,\ldots,\psi_{k-m+1})^\mathrm{T} and mm is the embedding dimension. For m=6m=6, the preceding analysis predicts that the embedded dynamics are Markovian. For a Markov process, the past and future are conditionally independent given the present, so Im=0I_m=0. A positive ImI_m indicates history dependence. To assess whether the embedded trajectories are consistent with the Markov property, we generate Markov trajectories by resampling and compare their CMI with that of the embedded trajectories. During resampling, we randomly reassign the values preceding similar embedded states while retaining each embedded state and its subsequent value. The reference range spans the minimum and maximum CMI estimates from the resampled Markov trajectories. Because of finite-sample effects, the CMI estimates for the Markov trajectories fluctuate around zero.

As ρN\rho_N varies, the six-dimensional estimated CMI I6I_6 for the embedded trajectories remains close to zero and lies within the reference ranges of the Markov trajectories [Figure S8]. Near ρc\rho_c, the estimated CMI ImI_m decreases with increasing embedding dimension mm and becomes comparable to that of the Markov trajectories at six dimensions [Figure S8]. These results support the Markov property of the six-dimensional delay-embedded dynamics.

Figure S8. CMI tests at τ=0.05\tau= 0.05 and Δt=0.01\Delta t = 0.01. (a) Six-dimensional CMI I6I_6 versus density. (b) CMI ImI_m versus embedding dimension mm near ρN=ρc\rho_N = \rho_c. Circles and squares denote CMI estimates from embedded trajectories and locally resampled Markov trajectories, respectively. Gray error bars show the reference ranges.

Landscape–flux theory and Jacobian spectra

Landscape, flux, and force decomposition

The nonequilibrium landscape-flux theory has been widely used to study nonequilibrium systems, with applications in many areas [5, 6, 7, 8, 9, 10].

We consider a stochastic system whose state vector x\mathbf{x} evolves according to the following Langevin equation:

dxdt=F(x)+ζ(x,t),(39)\frac{\mathrm{d}\mathbf{x}}{\mathrm{d}t}=\mathbf{F}(\mathbf{x})+\boldsymbol{\zeta}(\mathbf{x},t), \tag*{(39)}

where F\mathbf{F} is the deterministic driving force and ζ\boldsymbol{\zeta} is the stochastic force with ⟨ζ(x,t)ζT(x,t′)⟩=2D(x)δ(t−t′)\langle\boldsymbol{\zeta}(\mathbf{x},t)\boldsymbol{\zeta}^{\mathrm{T}}(\mathbf{x},t')\rangle=2\mathbf{D}(\mathbf{x})\delta(t-t'), where D(x)\mathbf{D}(\mathbf{x}) is the diffusion tensor. For the delay-embedded dynamics, the state vector is x=(x0,…,xnτ)T\mathbf{x}=(x_0,\ldots,x_{n_\tau})^{\mathrm{T}}, with components xr=ψ(t−rΔt)x_r=\psi(t-r\Delta t) for r=0,…,nτr=0,\ldots,n_\tau. Owing to the equivalence between the Langevin and Fokker–Planck descriptions, Eq. (39) can be rewritten as the Fokker–Planck equation for the probability distribution P(x,t)P(\mathbf{x},t):

∂P(x,t)∂t=−∇⋅J(x,t),(40)\frac{\partial P(\mathbf{x},t)}{\partial t}=-\nabla\cdot\mathbf{J}(\mathbf{x},t), \tag*{(40)}

where J\mathbf{J} is the nonequilibrium probabilistic flux. At the nonequilibrium steady state, the distribution becomes time independent, so that ∇⋅Jss=0\nabla\cdot\mathbf{J}_{\mathrm{ss}}=0, where the subscript ss denotes the steady state. In contrast to the vanishing flux at equilibrium, the nonequilibrium steady state only requires Jss\mathbf{J}_{\mathrm{ss}} to be divergence-free, and hence Jss\mathbf{J}_{\mathrm{ss}} can persist as a rotational field. From the Fokker–Planck equation, Jss\mathbf{J}_{\mathrm{ss}} reads:

Jss(x)=F(x)Pss(x)−∇⋅(DPss(x)).(41)\mathbf{J}_{\mathrm{ss}}(\mathbf{x})=\mathbf{F}(\mathbf{x})P_{\mathrm{ss}}(\mathbf{x})-\nabla\cdot\left(\mathbf{D}P_{\mathrm{ss}}(\mathbf{x})\right). \tag*{(41)}

In analogy with the equilibrium potential, the nonequilibrium potential landscape in the steady state is defined as

Uneq(x)=−ln⁡Pss(x).(42)U_{\mathrm{neq}}(\mathbf{x})=-\ln P_{\mathrm{ss}}(\mathbf{x}). \tag*{(42)}

Different from the equilibrium case, in which the driving force derives solely from the potential gradient, F\mathbf{F} can be decomposed into three parts:

F(x)=JssPss−D∇Uneq+∇⋅D=Fcurl+Fgrad+FD,(43)\mathbf{F}(\mathbf{x})=\frac{\mathbf{J}_{\mathrm{ss}}}{P_{\mathrm{ss}}}-\mathbf{D}\nabla U_{\mathrm{neq}}+\nabla\cdot\mathbf{D}=\mathbf{F}_{\mathrm{curl}}+\mathbf{F}_{\mathrm{grad}}+\mathbf{F}_{D}, \tag*{(43)}

where Fcurl:=Jss/Pss\mathbf{F}_{\mathrm{curl}}:=\mathbf{J}_{\mathrm{ss}}/P_{\mathrm{ss}} is the curl component of the driving force (a rotational flux force), Fgrad:=−D∇Uneq\mathbf{F}_{\mathrm{grad}}:=-\mathbf{D}\nabla U_{\mathrm{neq}} is the gradient force of the potential landscape, and FD:=∇⋅D\mathbf{F}_{D}:=\nabla\cdot\mathbf{D} is the diffusion force arising from the spatially dependent noise. Note that the curl force equals the probability velocity field vss\mathbf{v}_{\mathrm{ss}} (computed via central differences), i.e., Fcurl=vss:=E[x(t+Δt)−x(t−Δt)2Δt∣x(t)=x]\mathbf{F}_{\mathrm{curl}}=\mathbf{v}_{\mathrm{ss}}:=\mathbb{E}\left[\left.\frac{\mathbf{x}(t+\Delta t)-\mathbf{x}(t-\Delta t)}{2\Delta t}\right|\mathbf{x}(t)=\mathbf{x}\right], so that Jss=Pssvss\mathbf{J}_{\mathrm{ss}}=P_{\mathrm{ss}}\mathbf{v}_{\mathrm{ss}}. For notational simplicity, we henceforth drop the subscripts on J\mathbf{J}, v\mathbf{v}, PP, and UU (the subscript ss for the first three and neq for UU) in the following and in the main text.

In the main text, we take τ=0.05\tau=0.05 and Δt=0.01\Delta t=0.01 as an example to show the anomalous behavior of the delayed CAP system. The corresponding delay-embedded state is x=(x0,…,x5)T\mathbf{x}=(x_0,\ldots,x_5)^{\mathrm{T}}, where xr=ψ(t−rΔt)x_r=\psi(t-r\Delta t) for r=0,…,5r=0,\ldots,5. To visualize the six-dimensional landscape and flux, we project onto the two-dimensional plane y=(x5,x0)T=(ψ(t−τ),ψ(t))T\mathbf{y}=(x_5,x_0)^{\mathrm{T}}=(\psi(t-\tau),\psi(t))^{\mathrm{T}} and integrate over the remaining coordinates z=(x1,x2,x3,x4)T\mathbf{z}=(x_1,x_2,x_3,x_4)^{\mathrm{T}}. Hence, the projected landscape and flux are

U2(y)=−ln⁡P2(y),P2(y)=∫P6(y,z) dz,J2(y)=∫(J6,5(y,z)J6,0(y,z)) dz.(44)\begin{aligned} U_2(\mathbf{y})&=-\ln P_2(\mathbf{y}),\qquad P_2(\mathbf{y})=\int P_6(\mathbf{y},\mathbf{z})\,\mathrm{d}\mathbf{z},\\ \mathbf{J}_2(\mathbf{y})&=\int\begin{pmatrix}J_{6,5}(\mathbf{y},\mathbf{z})\\J_{6,0}(\mathbf{y},\mathbf{z})\end{pmatrix}\,\mathrm{d}\mathbf{z}. \tag*{(44)} \end{aligned}

Here, P6(x)P_6(\mathbf{x}) and J6(x)=(J6,0(x),…,J6,5(x))T\mathbf{J}_6(\mathbf{x})=(J_{6,0}(\mathbf{x}),\ldots,J_{6,5}(\mathbf{x}))^{\mathrm{T}} denote the probability density and flux in the six-dimensional delay-embedded space, respectively.

To quantify the global circulation of the flux, we define the mean flux as the spatial average of ∣J∣\lvert\mathbf{J}\rvert over the delay-embedded domain Ω\Omega:

Ja:=⟨∣J∣⟩=1∣Ω∣∫Ω∣J∣ dx.(45)J_a:=\langle\lvert\mathbf{J}\rvert\rangle=\frac{1}{\lvert\Omega\rvert}\int_{\Omega}\lvert\mathbf{J}\rvert\,\mathrm{d}\mathbf{x}. \tag*{(45)}

The mean flux measures the average strength of the flux field. In the main text, JaJ_a is computed in the 6-dimensional delay-embedded space for each neighborhood density ρN\rho_N, and its density slope Ja′:=d⟨∣J∣⟩/dρNJ'_a:=\mathrm{d}\langle\lvert\mathbf{J}\rvert\rangle/\mathrm{d}\rho_N provides a dynamical signature of the bifurcation.

Jacobian spectra and stability

For nτ=5n_{\tau}=5, the linearized delay-difference equation (S8) for perturbations about ψ∗=0\psi^{*}=0 can be written in the delay-embedded space as

uk+1=uk+Δt AFuk=Muk,(46)\mathbf{u}_{k+1}=\mathbf{u}_{k}+\Delta t\,A_{F}\mathbf{u}_{k}=M\mathbf{u}_{k}, \tag*{(46)}

where uk=(uk,uk−1,…,uk−5)T\mathbf{u}_{k}=(u_{k},u_{k-1},\ldots,u_{k-5})^{\mathrm{T}} and

M:=(10000−βΔt100000010000001000000100000010),AF:=Jac⁡F(0)=M−IΔt.(47)M:=\begin{pmatrix} 1 & 0 & 0 & 0 & 0 & -\beta\Delta t \\ 1 & 0 & 0 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 0 & 0 & 0 \\ 0 & 0 & 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & 0 & 1 & 0 \end{pmatrix}, \qquad A_{F}:=\operatorname{Jac}F(0)=\frac{M-\mathbb{I}}{\Delta t}. \tag*{(47)}

Here, I\mathbb{I} denotes the 6×66\times6 identity matrix.

Let aja_j denote the Jacobian eigenvalues of the driving force averaged over 1Δt1\Delta t, and μj\mu_j the eigenvalues of the one-step map MM. In the corresponding continuous-time linear dynamics, the transient driving force has Jacobian Ainst:=log⁡M/ΔtA_{\mathrm{inst}}:=\log M/\Delta t. Its eigenvalues λj\lambda_j satisfy

μj=1+Δt aj=eλjΔt,Re⁡λj=ln⁡∣μj∣Δt.(48)\mu_j=1+\Delta t\,a_j=e^{\lambda_j\Delta t}, \qquad \operatorname{Re}\lambda_j=\frac{\ln|\mu_j|}{\Delta t}. \tag*{(48)}

At the bifurcation, max⁡jRe⁡λj\max_j\operatorname{Re}\lambda_j crosses zero from below, indicating that the fixed point loses stability.

Applying the same construction to each component force in Eq. (43), we obtain the Jacobian eigenvalues λq,j\lambda_{q,j} of the corresponding transient force, where q∈{curl,grad,D}q\in\{\mathrm{curl},\mathrm{grad},D\}. Near the bifurcation, max⁡jRe⁡λq,j\max_j\operatorname{Re}\lambda_{q,j} remains positive for the curl force, negative for the gradient force, and slightly positive for the diffusion force, as shown in Figure S9. Thus, the potential landscape tends to stabilize the fixed point, whereas the flux tends to destabilize it. Together, these results support the flux as the dynamical origin of the anomalous DPR.

Figure S9. Largest real parts of the Jacobian eigenvalues of the transient curl (green), gradient (orange), total (blue), and diffusion (purple) forces at the six-dimensional origin, for τ=0.05\tau=0.05 and Δt=0.01\Delta t=0.01. The red line marks ρc\rho_c.

References

  1. [1]T. D. Frank, P. J. Beek, and R. Friedrich. Fokker–Planck perspective on stochastic delay systems: Exact solutions and data analysis of biological systems. Physical Review E, 68(2):021912, 2003.pubmed.ncbi.nlm.nih.gov/14525011
  2. [2]Uwe Küchler and Beatrice Mensch. Langevins stochastic differential equation extended by a time-delayed term. Stochastics and Stochastic Reports, 40(1–2):23–42, 1992.DOI
  3. [3]Carlos A. Lugo and Alan J. McKane. Quasicycles in a spatial predator-prey model. Physical Review E, 78(5):051911, 2008.pubmed.ncbi.nlm.nih.gov/19113159
  4. [4]A. J. McKane and T. J. Newman. Predator-prey cycles from resonant amplification of demographic stochasticity. Physical Review Letters, 94(21):218102, 2005.DOI
  5. [5]Jin Wang, Li Xu, and Erkang Wang. Potential landscape and flux framework of nonequilibrium networks: Robustness, dissipation, and coherence of biochemical oscillations. Proceedings of the National Academy of Sciences, 105(34):12271–12276, 2008.DOI
  6. [6]Jin Wang. Landscape and flux theory of non-equilibrium dynamical systems with application to biology. Advances in Physics, 64(1):1–137, 2015.DOI
  7. [7]Xiaona Fang, Karsten Kruse, Ting Lu, and Jin Wang. Nonequilibrium physics in biology. Reviews of Modern Physics, 91(4):045004, 2019.DOI
  8. [8]Jie Su, Zhiyu Cao, Jin Wang, Huijun Jiang, and Zhonghuai Hou. Dynamical and thermodynamical origins of motility-induced phase separation. Cell Reports Physical Science, 5(2):101817, 2024.DOI
  9. [9]Ruizhe Yan, Jie Su, and Jin Wang. Mechanism of the nonequilibrium phase transition in self-propelled particles with alignment. Physical Review Research, 7(3):033082, 2025.arxiv.org/abs/2411.06818
  10. [10]Dongrun Jian, Jie Su, Jun Wang, and Jin Wang. Uncovering the underlying mechanisms of phase transitions in chiral active particles. Communications Physics, 8(1):386, 2025.arxiv.org/abs/2503.18388

Paper details

vlnc:W_ftqbjcc8
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents