betaFeedback

WorksPhysics and Astronomy

Effective Temperature and Einstein Relation for Particles in Mesoscale Turbulence

Abstract

From the smallest scales of quantum systems to the largest scales of intergalactic medium, turbulence is ubiquitous in nature. Often dubbed as the last unsolved problem of classical physics, it remains a time tested paradigm of dynamics far from equilibrium. The phenomenon even transcends to self-propelled fluids such as dense bacterial suspensions that can display turbulence at mesoscale even though the constituent particles move at Reynolds number below unity. It is intensely debated whether such fluids possess an effective temperature and obey fluctuation-dissipation relations (FDR) as they are generally marred by a lack of detailed balance. In this letter, we answer this question and report an exact expression of the effective temperature for a distribution of interacting particles that are advected by a mesoscale turbulent flow. This effective temperature is linear in particle diffusivity with the slope defining the particle mobility that is higher when the background fluid exhibits global polar ordering, and lower when the fluid is in isotropic equilibrium. We believe our work is a direct verification of the Einstein relation -the simplest FDR, for interacting particles immersed in a mesoscale turbulence.

Fluid turbulence has been fascinating to us since time immemorial. From the deluge drawings of Leonardo da Vinci [1] dating almost five centuries ago to the relatively modern works on complex fluids [2], plasmas [3] and atmospheric systems [4], fluid turbulence is ubiquitous in nature. Despite many decades of exhaustive research, this chaotic state of fluid motion continues to remain as the oldest unsolved problem in classical physics. Put simply, there is currently no rigorous theory that starts from the Navier-Stokes equation and leads to the experimentally observed laws on energy dissipation and velocity structure functions in turbulent flows [5, 6]. Nevertheless, turbulence is considered as a classical paradigm of nonlinear dynamics far from thermodynamic equilibrium [7–9]. An exciting area of mesoscale turbulence has recently emerged where the fluid displays spatiotemporal chaos even though its constituent units are self-propelled and move at Reynolds number below unity. Examples include dense bacterial suspensions [10, 11], microtubule networks [12], artificial swimmers [13] and active liquid crystals [14], to mention a few. It is a matter of intense debate whether these self-propelled fluids should satisfy fluctuation-dissipation relations (FDR) as the constituent particle dynamics generally lack detailed balance [15, 16]. A related and important question is whether one can describe an effective temperature of particles dispersed in such fluids [17, 18]. In this report, we address these fundamental issues by predicting an exact expression of the effective temperature for a distribution of interacting particles that are advected by an active turbulent flow in the background. This effective temperature is linear in particle diffusivity with the slope characterizing the particle mobility, that is higher when the background fluid exhibits global polar ordering, and lower when the fluid is in isotropic equilibrium. Our work is therefore a direct verification of the celebrated Einstein relation - the simplest FDR, for interacting particles advected in mesoscale turbulence. The results reported here

are valid across four decades of variation in the damping coefficient, thereby putting a large number of active systems within the ambit of our work—from dense suspensions of microswimmers that are traditionally overdamped [19, 20, 21] to the lesser understood underdamped suspensions where inertial effects are significant, notable examples being ciliates, planktons, copepods and other macroswimmers moving in a background media of low viscosity [22]. In what follows, we provide the details of the phenomenological model and methods used in our work.

Model/Methods: To mimic a generic active suspension, we use a minimal continuum model of an incompressible active fluid in two dimensions [23, 24], whose streamfunction Ψ(x,y)\Psi(x,y) evolves according to the following equation

∂∂t(∇2Ψ)+λ0∂(Ψ,∇2Ψ)∂(x,y)=−Γ0∇4Ψ−Γ2∇6Ψ−μ∇2Ψ(1)\frac{\partial}{\partial t}\left(\nabla^{2}\Psi\right)+\lambda_{0}\frac{\partial\left(\Psi,\nabla^{2}\Psi\right)}{\partial(x,y)}=-\Gamma_{0}\nabla^{4}\Psi-\Gamma_{2}\nabla^{6}\Psi-\mu\nabla^{2}\Psi \tag*{(1)}

here the non-dimensional parameter λ0\lambda_{0} decides the type of active unit, meaning they are either pusher or a puller, corresponding respectively to the case λ0>0\lambda_{0}>0 or λ0<0\lambda_{0}<0. The model was used recently to investigate pattern formation [25] and transport coefficients in dense active liquids [26]. We set the value of λ0=3.5\lambda_{0}=3.5 throughout our work implying that we have a pusher type of active units. We keep Γ0,2>0\Gamma_{0,2}>0 to enable energy injection into the active liquid via fluid instabilities. The scalar field μ=α+β∣u∣2\mu=\alpha+\beta|\mathbf{u}|^{2} depends on the local velocity u=∇×Ψz^\mathbf{u}=\nabla\times\Psi\hat{\mathbf{z}}, and was first introduced by Toner and Tu to model the “flocking” behavior in self-propelled rod-like objects [28, 29]. The parameter α\alpha, henceforth referred to as the Ekman friction, acts at intermediate scales and can either lead to a damping of energy when α>0\alpha>0 or an injection of energy when α<0\alpha<0. Former leads the fluid to an isotropic equilibrium and the latter yields a globally ordered polar state with mean velocity ∣α∣/β\sqrt{|\alpha|/\beta}. To non-dimensionalize eq. (1), we normalize all distances to σ0=5π2Γ2/Γ0\sigma_0 = 5\pi\sqrt{2\Gamma_2/\Gamma_0} and all times to t0=5π2Γ2/Γ02t_0 = 5\pi\sqrt{2\Gamma_2/\Gamma_0^2}. In terms of these reduced units, we fix the values of the model parameters as Γ0=(5π2)−1\Gamma_0 = (5\pi\sqrt{2})^{-1}, Γ2=(5π2)−3\Gamma_2 = (5\pi\sqrt{2})^{-3} and β=0.5\beta= 0.5, in order to remain consistent with earlier works [26, 30]. Equation (1) is then numerically solved using a pseudo-spectral approach over a square grid of 5122512^2 points in a doubly periodic box of size 2π2\pi [31]. A time step of Δt=2×10−4\Delta t = 2 \times10^{-4} is employed for time marching Ψ\Psi with the Crank-Nicholson scheme that is sufficient to maintain numerical stability in the entire range of parameters explored here.

Into the active turbulent flow, we throw NN particles that follow the dynamics:

dridt=vidvidt=−γ(vi−u(ri))−∇iΦ(2)\begin{aligned} \frac{\mathrm{d}\boldsymbol{r}_i}{\mathrm{d}t} &= \boldsymbol{v}_i \\ \frac{\mathrm{d}\boldsymbol{v}_i}{\mathrm{d}t} &= -\gamma\left(\boldsymbol{v}_i-\boldsymbol{u}(\boldsymbol{r}_i)\right)-\nabla_i\Phi \tag*{(2)} \end{aligned}

where γ\gamma is the damping coefficient and (ri,vi)(\boldsymbol{r}_i,\boldsymbol{v}_i) correspond respectively to position and velocity of an ithi^{\mathrm{th}} particle. We realize u(ri)\boldsymbol{u}(\boldsymbol{r}_i) as the turbulent flow field projected at the location of this particle, and −∇iΦ-\nabla_i\Phi as the mass normalized force acting on it. The limit γ⟶∞\gamma\longrightarrow\infty naturally corresponds to an overdamped dynamics of passive tracers advected by a background flow. In this limit, the tracers can exhibit an intervening anomalous diffusive regime as the Ekman friction is pushed to negatively larger values, say α<−6\alpha< -6 [27]. The limit γ⟶0\gamma\longrightarrow0 on the other hand, corresponds to the Newtonian dynamics of particles without any background fluid. At intermediate values of γ\gamma, both inertial and damping effects coexist thereby rendering generality to the particle dynamics. We set the interaction between particles as the repulsive Weeks-Chandler-Andersen potential energy Φ=∑j<kϕ(∣rj−rk∣)\Phi= \sum_{j<k}\phi(\lvert\boldsymbol{r}_j-\boldsymbol{r}_k\rvert), where the potential

ϕ(r)={4ϵ[(σ0r)12−(σ0r)6]+ϵ,r<21/6σ00,r>21/6σ0(3)\phi(r)= \begin{cases} 4\epsilon\left[\left(\frac{\sigma_0}{r}\right)^{12}-\left(\frac{\sigma_0}{r}\right)^6\right]+\epsilon, & r<2^{1/6}\sigma_0 \\ 0, & r>2^{1/6}\sigma_0 \end{cases} \tag*{(3)}

All measurements on the particles as well as the turbulent fluid are carried out only after the latter attains a steady state. To improve our statistics, we average these measurements over 100 independently prepared realizations of the initial state. Below we show that the statistics of the Eulerian velocity u\boldsymbol{u} is indeed profitable to the prediction of an effective temperature of our particles.

Statistics of u\boldsymbol{u}: We start with a measurement of the Eulerian flow field projected at the particle positions, i.e u(ri)\boldsymbol{u}(\boldsymbol{r}_i). In Figure 1, we show the distribution of a component of this velocity in the steady state and realize that it is well approximated by a Gaussian. In the inset, we show its temporal auto-correlation that is clearly an exponential with a relaxation time τ\tau. Combining these two facts, we can claim that the fluid velocity projected at particle locations is a Gaussian colored noise of zero mean and variance given as -

Figure 1. Distribution of Eulerian velocities projected on the particles fits well to a Gaussian. Data shown here corresponds to α=3\alpha= 3; results are similar for other α\alpha. Inset: Temporal auto-correlation of u\boldsymbol{u} with a relaxation time τ\tau. We can thus visualize u\boldsymbol{u} as a Gaussian colored noise - a headway that will be used to formulate an effective temperature for our particles.

⟨uiχ(t)ujψ(t′)⟩=12δijδχψurms2exp⁡(−∣t−t′∣/τ)(4)\left\langle u_{i\chi}(t)u_{j\psi}(t')\right\rangle= \frac{1}{2}\delta_{ij}\delta_{\chi\psi}u_{\mathrm{rms}}^2\exp\left(-\lvert t-t'\rvert/\tau\right) \tag*{(4)}

where urmsu_{\mathrm{rms}} is a velocity scale that sets the kinetic energy of the background fluid in the steady state. We use the Latin symbols i,ji,j to denote particle labels and the Greek symbols χ,ψ\chi,\psi to denote spatial indices. In the light of all these arguments, we realize that the stochastic dynamics in Eq. (2) is essentially non-Markovian and its corresponding steady state Koffler-Planck equation must be carefully solved (see next) to derive an exact expression of the stationary probability distribution. Factorization of this distribution into position and momentum parts will yield an exact expression of the effective temperature of our particles. This is done next.

Theoretical Prediction of TeffT_{\mathrm{eff}}: To systematically derive an expression for the effective temperature TeffT_{\mathrm{eff}} in terms of the fluid properties, we will now set up a Fokker-Planck equation for the governing dynamics of the interacting particles and solve it to get a stationary velocity distribution. As stated earlier, the dynamics shown in Eq. (2) is essentially non-Markovian due to the Gaussian colored noise u\boldsymbol{u}. An exact solution of the corresponding Fokker-Planck equation, with the exception of harmonic interactions, is intractable in the general case and approximations are always required [32]. To circumvent this difficulty, we reduce the original non-Markovian dynamics in Eq. (2) to an equivalent Markovian dynamics by an appropriate enlargement of the phase space [33, 34].

Figure 2. Evidence of linear relationship between effective temperature TeffT_{\mathrm{eff}} and particle diffusivity DD is presented here at various value of damping coefficient γ\gamma. The numerical value of TeffT_{\mathrm{eff}} is extracted by fitting u\mathbf{u} to a Gaussian and the theoretical value is predicted by Eq. (9). Dashed line is a linear fit to our numerical data and has a slope that indicates the inverse of particle mobility μ\mu. This is a direct verification of the Einstein’s relation, the simplest FDR. Particle diffusivity D=lim⁡t→∞⟨Δr2⟩/(4t)D=\lim_{t\to\infty}\langle\Delta r^{2}\rangle/(4t) is varied by varying the fluid friction α\alpha. For each panel that corresponds to a specific γ\gamma, the mobility of particles is higher when the background fluid is in a globally ordered polar state (α<0\alpha<0), and lower when it is in isotropic equilibrium (α>0\alpha>0).

In the component form, this simply reads

driχdt=viχdviχdt=−γ(viχ−uiχ)−∇iχΦduiχdt=−1τuiχ+urms2τξiχ(5)\begin{aligned} \frac{d r_{i\chi}}{dt} &= v_{i\chi}\\ \frac{d v_{i\chi}}{dt} &= -\gamma(v_{i\chi}-u_{i\chi})-\nabla_{i\chi}\Phi\\ \frac{d u_{i\chi}}{dt} &= -\frac{1}{\tau}u_{i\chi}+\sqrt{\frac{u_{\mathrm{rms}}^{2}}{\tau}}\xi_{i\chi} \tag*{(5)} \end{aligned}

where ξiχ\xi_{i\chi} is a Gaussian white noise with zero mean and correlation ⟨ξiχ(t)ξjψ(t′)⟩=δijδχψδ(t−t′)\langle\xi_{i\chi}(t)\xi_{j\psi}(t')\rangle=\delta_{ij}\delta_{\chi\psi}\delta(t-t'). The reader should note that the dynamics in Eq. (5) is consistent with the expected correlation properties of u\mathbf{u} that we already know from Eq. (4). The Fokker-Planck equation for the probability distribution P(rχ,vχ,t)P(r_{\chi},v_{\chi},t) corresponding to the dynamics in (5) can be written using a decoupling approximation as [35]:

∂∂tP(rχ,vχ,t)=LP(rχ,vχ,t)(6)\frac{\partial}{\partial t}P(r_{\chi},v_{\chi},t)=\mathcal{L}P(r_{\chi},v_{\chi},t) \tag*{(6)}

where the operator L\mathcal{L} is defined as

L=[−vχ∂∂rχ+∂∂vχ(γvχ+∂Φ∂rχ)+γ2urms2τ2(1+γτ+τ2⟨∂2Φ∂rχ2⟩)∂2∂vχ2+γ2urms2τ22(1+γτ+τ2⟨∂2Φ∂rχ2⟩)∂2∂rχ∂vχ],(7)\mathcal{L}=\left[-v_{\chi}\frac{\partial}{\partial r_{\chi}}+\frac{\partial}{\partial v_{\chi}}\left(\gamma v_{\chi}+\frac{\partial\Phi}{\partial r_{\chi}}\right)+\frac{\gamma^{2}u_{\mathrm{rms}}^{2}\tau}{2\left(1+\gamma\tau+\tau^{2}\left\langle\frac{\partial^{2}\Phi}{\partial r_{\chi}^{2}}\right\rangle\right)}\frac{\partial^{2}}{\partial v_{\chi}^{2}}+\frac{\gamma^{2}u_{\mathrm{rms}}^{2}\tau^{2}}{2\left(1+\gamma\tau+\tau^{2}\left\langle\frac{\partial^{2}\Phi}{\partial r_{\chi}^{2}}\right\rangle\right)}\frac{\partial^{2}}{\partial r_{\chi}\partial v_{\chi}}\right], \tag*{(7)}

The steady state solution of the above Fokker-Planck equation readily factorizes into position and velocity parts as:

Ps(rχ,vχ)=1Zexp⁡(−vχ22σχ2)exp⁡[−Φσχ2(1+γτ)](8)P_s(r_\chi,v_\chi)=\frac{1}{Z}\exp\left(-\frac{v_\chi^2}{2\sigma_\chi^2}\right)\exp\left[-\frac{\Phi}{\sigma_\chi^2(1+\gamma\tau)}\right] \tag*{(8)}

with the velocity variance realized as

σχ2=urms2γτ2(1+γτ+τ2⟨∂2Φ∂xχ2⟩)≡Teff(9)\sigma_\chi^2=\frac{u_{\mathrm{rms}}^2\gamma\tau}{2\left(1+\gamma\tau+\tau^2\left\langle\frac{\partial^2\Phi}{\partial x_\chi^2}\right\rangle\right)}\equiv T_{\mathrm{eff}} \tag*{(9)}

the effective temperature of our inertial particles, with χ\chi being xx or yy. We take a moment of pause to appreciate the beauty of this exact expression. There are no fitting parameters here and we are able to completely predict the effective temperature of particles once the background fluid properties (urmsu_{\mathrm{rms}}, τ\tau, γ\gamma) and particle interactions (Φ\Phi) are specified. The predicted TeffT_{\mathrm{eff}} is in excellent agreement with the numerically estimated value that is obtained by fitting the velocity distribution to a Gaussian. This is shown in Fig. 2 where we plot TeffT_{\mathrm{eff}} vs. DD, the particle diffusivity. For each γ\gamma, we calculate the diffusivity as D=lim⁡t→∞⟨Δr2⟩/(4t)D=\lim_{t\to\infty}\langle\Delta r^2\rangle/(4t) and its variation Dmin⁡→Dmax⁡D_{\min}\to D_{\max} comes from varying the fluid friction α\alpha in the range 4.0→−4.04.0\to-4.0. It is evident from each panel of this figure that TeffT_{\mathrm{eff}} is linear in DD with a slope yielding the inverse of particle mobility μ\mu. This is a direct verification of the Einstein’s relation, the simplest formulation of FDR. We also observe that the particle mobility is higher when the background fluid is in a globally ordered polar state, i.e α<0\alpha<0, and lower when the fluid is in isotropic equilibrium, or α>0\alpha>0. Our results are valid across four decades of variation in damping coefficient effectively covering the entire range 1<γτ<1041<\gamma\tau<10^4, where the lower and upper limit corresponding respectively to the inertial and overdamped regime. This puts a large number of active systems within the purview of our work—from overdamped suspensions of microswimmers [36, 37] to the underdamped suspensions of macroswimmers such as ciliates, planktons, copepods moving in low viscosity media [22].

Acknowledgments

The authors thank V Balakrishnan, Sumesh Thampi and Abhijit Sen for discussions and comments on the manuscript. All simulations were done from the HPC-Physics cluster of our group. Support from the SERB core grant, DST and New Faculty Seed Grant (NFSG), IIT Madras, is gratefully acknowledged.

References

References

  1. [15]L. Caprini, A. Puglisi, and A. Sarracino, Fluctuation–dissipation relations in active matter systems, Symmetry 13, 10.3390/sym13010081 (2021).arxiv.org/abs/2012.06427
  2. [16]D. Martin, J. O’Byrne, M. E. Cates, E. Fodor, C. Nardini, J. Tailleurr, and F. van Wijland, Statistical mechanics of active ornstein-uhlenbeck particles, Phys. Rev. E 103, 032607 (2021).DOI
  3. [17]L. F. Cugliandolo, The effective temperature, Journal of Physics A: Mathematical and Theoretical 44, 483001 (2011).DOI
  4. [18]A. Puglisi, A. Sarracino, and A. Vulpiani, Temperature in and out of equilibrium: A review of concepts, tools and attempts, Physics Reports 709-710, 1 (2017), temperature in and out of equilibrium: a review of concepts, tools and attempts.DOI
  5. [19]G. Gompper, R. G. Winkler, T. Speck, A. Solon, C. Nardini, F. Peruani, H. Löwen, R. Golestanian, U. B. Kaupp, L. Alvarez, T. Kiørboe, E. Lauga, W. C. K. Poon, A. DeSimone, S. Muiños-Landin, A. Fischer, N. A. Söker, F. Cichos, R. Kapral, P. Gaspard, M. Ripoll, F. Sagües, A. Doostmohammadi, J. M. Yeomans, I. S. Aranson, C. Bechinger, H. Stark, C. K. Hemelrijk, F. J. Nedelec, T. Sarkar, T. Aryaksama, M. Lacroix, G. Duclos, V. Yashunsky, P. Silberzan, M. Arroyo, and S. Kale, The 2020 motile active matter roadmap, Journal of Physics: Condensed Matter 32, 193001 (2020).DOI
  6. [20]M. C. Marchetti, J. F. Joanny, S. Ramaswamy, T. B. Liverpool, J. Prost, M. Rao, and R. A. Simha, Hydrodynamics of soft active matter, Rev. Mod. Phys. 85, 1143 (2013).DOI
  7. [21]C. Bechinger, R. Di Leonardo, H. Löwen, C. Reichhardt, G. Volpe, and G. Volpe, Active particles in complex and crowded environments, Rev. Mod. Phys. 88, 045006 (2016).DOI
  8. [22]D. Klotsa, As above, so below, and also in between: mesoscale active matter in fluids, Soft Matter 15, 8946 (2019).DOI
  9. [23]H. H. Wensink, J. Dunkel, S. Heidenreich, K. Drescher, R. E. Goldstein, H. Löwen, and J. M. Yeomans, Mesoscale turbulence in living fluids, Proceedings of the National Academy of Sciences 109, 14308 (2012).
  10. [24]J. Dunkel, S. Heidenreich, M. Bär, and R. E. Goldstein, Minimal continuum theories of structure formation in dense active fluids, New Journal of Physics 15, 045016 (2013).DOI
  11. [25]M. James, W. J. T. Bos, and M. Wilczek, Turbulence and turbulent pattern formation in a minimal model for active fluids, Phys. Rev. Fluids 3, 061101 (2018).DOI
  12. [26]S. C. P. and A. Joy, Friction scaling laws for transport in active turbulence, Phys. Rev. Fluids 5, 024302 (2020).
  13. [27]S. Mukherjee, R. K. Singh, M. James, and S. S. Ray, Anomalous diffusion and lévy walks distinguish active from inertial turbulence, Physical Review Letters 127, 118001 (2021).DOI
  14. [28]J. Toner, Y. Tu, and S. Ramaswamy, Hydrodynamics and phases of flocks, Annals of Physics 318, 170 (2005).DOI
  15. [29]J. Toner and Y. Tu, Flocks, herds, and schools: A quantitative theory of flocking, Physical review E 58, 4828 (1998).DOI
  16. [30]V. Bratanov, F. Jenko, and E. Frey, New class of turbulence in active fluids, Proceedings of the National Academy of Sciences 112, 15048 (2015).DOI
  17. [31]C. Canuto, M. Y. Hussaini, A. Quarteroni, A. Thomas Jr, et al., Spectral methods in fluid dynamics (Springer Science & Business Media, 2012).
  18. [32]P. Hanggi and P. Jung, Colored Noise in Dynamical Systems, Advances in Chemical Physics 89, 239 (1995).
  19. [33]N. G. Van Kampen, Stochastic processes in physics and chemistry, Vol. 1 (Elsevier, 1992).
  20. [34]H. Risken, Fokker-planck equation, in The Fokker-Planck Equation (Springer, 1996), pp. 63–95.
  21. [35]L. Fronzoni, P. Grigolini, P. Hanggi, F. Moss, R. Mannella, and P. V. E. McClintock, Bistable oscillator dynamics driven by nonwhite noise, Phys. Rev. A 33, 3320 (1986).pubmed.ncbi.nlm.nih.gov/9897041
  22. [36]X.-L. Wu and A. Libchaber, Particle diffusion in a quasi-two-dimensional bacterial bath, Physical review letters 84, 3017 (2000).pubmed.ncbi.nlm.nih.gov/11019000
  23. [37]F. Kümmel, B. ten Hagen, R. Wittkowski, I. Buttinoni, R. Eichhorn, G. Volpe, H. Löwen, and C. Bechinger, Circular motion of asymmetric self-propelling particles, Phys. Rev. Lett. 110, 198302 (2013).

Paper details

vlnc:W_ucm8waqd
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