betaFeedback

WorksPhysics and Astronomy

A fluctuating lattice Boltzmann formulation based on orthogonal central moments

Abstract

Thermal fluctuations play a central role in fluid dynamics at mesoscopic scales and must be incorporated into numerical schemes in a manner consistent with statistical mechanics. In this work, we develop a fluctuating lattice Boltzmann formulation based on an orthogonal central-moments-based representation. Stochastic forcing is introduced directly in the space of central moments (CMs) and consistently paired with mode-dependent relaxation, yielding a discrete kinetic model that satisfies the fluctuation-dissipation theorem exactly at the lattice level. Owing to the orthogonality of the basis, the equilibrium covariance matrix of the central moments is diagonal, and each non-conserved mode can be interpreted as an independent discrete Ornstein-Uhlenbeck process with variance fixed by equilibrium thermodynamics. The resulting formulation guarantees exact equipartition of kinetic energy at equilibrium, preserves Galilean invariance, and retains the enhanced numerical stability characteristic of CMs-based collision operators. Explicit fluctuating schemes are constructed for the D2Q9 and D3Q27 lattices. The extension to reduced-velocity discretisation is discussed too. A comprehensive set of numerical tests verifies correct thermalisation, isotropy of equilibrium statistics, and the expected scaling of velocity fluctuations with thermal energy, density, and relaxation time. In contrast to fluctuating BGK formulations, the present method remains stable and well posed in the over-relaxation regime, including in the immediate vicinity of the stability limit. These results demonstrate that CMs-based lattice Boltzmann methods provide a natural and robust framework for fluctuating hydrodynamics, in which dissipation, noise, and kinetic mode structure are consistently aligned at the discrete level.

Introduction

Thermal fluctuations are an intrinsic feature of fluid motion at mesoscopic and microscopic scales, where stochastic effects coexist with deterministic hydrodynamic transport[1]. In such regimes, fluctuations influence equilibrium properties, transport coefficients, and spatio-temporal correlations, and must be accounted for to obtain physically consistent descriptions. Numerical methods aiming to resolve these phenomena are therefore required to respect both hydrodynamic conservation laws and the fluctuation–dissipation theorem (FDT)[2].

The lattice Boltzmann method (LBM), derived from a discrete kinetic description, provides a natural framework for incorporating thermal fluctuations at the mesoscopic level[3]. In LBM, fluid motion is represented by collections of particle distribution functions (also known as populations) that stream and collide on a fixed Cartesian lattice[4]. By augmenting the kinetic dynamics with stochastic forcing, fluctuating lattice Boltzmann methods (FLBMs) are able to reproduce equilibrium velocity and stress fluctuations and recover fluctuating hydrodynamics in the macroscopic limit[5]. Over the past two decades, a variety of fluctuating LB formulations have been proposed, most of them based on single-relaxation-time or raw-moment multi-relaxation-time collision operators[6]. While successful in many applications, these approaches often suffer from limited robustness and degraded accuracy in regimes characterised by low viscosity, strong non-equilibrium effects, or stringent stability requirements. Indeed, the choice of collision operator plays a central role in this context. Single-relaxation-time (also know as BGK, that is the acronym for Bhatnagar–Gross–Krook) models[7] offer conceptual simplicity but provide limited control over non-hydrodynamic modes, whose uncontrolled dynamics may contaminate fluctuation spectra and compromise numerical stability. Multi-relaxation-time (MRT) formulations[8] introduce additional flexibility through mode-dependent relaxation, yet when expressed in terms of raw moments they may exhibit spurious couplings between hydrodynamic and higher-order modes, particularly at finite velocities. These limitations are amplified in fluctuating settings, where improper mode coupling or inconsistent noise allocation readily leads to violations of equipartition, distorted fluctuation spectra, and unphysical correlations[9].

A significant advance in this direction has been achieved very recently by Lauricella et al.[10], who introduced a regularised fluctuating lattice Boltzmann method based on a full Hermite expansion on the D3Q27 lattice. By reconstructing both equilibrium and non-equilibrium populations on an orthogonal Hermite basis and carefully controlling the relaxation of hydrodynamic and non-hydrodynamic (ghost) modes, this approach restores thermodynamic consistency and significantly improves numerical stability across a broad range of relaxation parameters. These results clearly demonstrate that accurate fluctuating hydrodynamics at the lattice level crucially relies on orthogonal mode decompositions and a clean separation between physical and non-physical degrees of freedom. This insight has been further sharpened in a sub-丁香五月 sequent work by the same authors[11], who proposed a ghost-mode filtered fluctuating lattice Boltzmann formulation. In this approach, ghost modes are prevented from carrying deterministic information altogether and are reduced to purely stochastic carriers of thermal noise, while hydrodynamic modes retain their physical dynamics. Remarkably, this minimalistic treatment yields fluctuation statistics comparable to fully regularised high-order schemes, underscoring that diagonal equilibrium covariance and strict mode decoupling are not merely advantageous, but essential for physically consistent fluctuating lattice Boltzmann formulations.

Central-moments-based lattice Boltzmann methods (CM-LBMs) provide a natural and robust framework in which these requirements can be satisfied intrinsically[12, 13]. By formulating the collision process in a reference frame moving with the local fluid velocity, such schemes improve Galilean invariance, suppress spurious velocity-dependent couplings, and significantly enhance numerical stability[14, 15]. These advantages have been demonstrated in a wide range of demanding applications, including turbulent, multiphase, and strongly non-equilibrium flows[16, 17]. Despite their favourable properties, the systematic construction of fluctuating lattice Boltzmann schemes within a central-moments framework has received comparatively limited attention.

Actually, incorporating thermal fluctuations into a CMs-based formulation poses specific challenges. Stochastic forcing must be introduced directly in moment space in a manner consistent with mode-dependent relaxation, while ensuring exact equilibrium statistics and strict compliance with the FDT. At the same time, the defining properties of central-moments schemes, including numerical robustness, must be preserved.

The aim of this work is to develop a fluctuating lattice Boltzmann formulation based on central moments that is both systematic and physically consistent. Stochastic collision operators are derived directly in space of central moments (CMs), with noise terms assigned mode by mode according to equilibrium statistical mechanics. The resulting formulation guarantees exact equipartition at equilibrium, recovers the correct fluctuating hydrodynamic behaviour in the macroscopic limit, and remains fully compatible with standard CMs bases and lattice discretisations.

The proposed approach retains the salient features of conventional CM-LBMs and is readily applicable to large-scale simulations[18]. Numerical tests are presented to assess equilibrium fluctuations, moment-wise energy partitioning, and robustness over a broad range of relaxation parameters. The results demonstrate that central-moments-based fluctuating lattice Boltzmann methods (CM-FLBM) provide accurate and stable descriptions of thermally fluctuating flows, including regimes in which traditional fluctuating BGK formulations become unreliable.

One of the key results of this work is that orthogonal central moments are not merely advantageous, but required to realise fluctuating hydrodynamics with diagonal equilibrium covariance and statistically independent stochastic modes at the lattice level. While correlated stochastic forcing could in principle restore the fluctuation–dissipation balance in non-orthogonal bases, the resulting noise covariance would depend explicitly on the local velocity, undermining Galilean invariance and obscuring the physical interpretation of individual modes. Unlike ghost-mode filtered formulations, which enforce statistical consistency by suppressing deterministic ghost-mode dynamics, the present approach retains a full central-moment kinetic structure and achieves diagonal equilibrium covariance intrinsically, without discarding deterministic degrees of freedom.

The specific contributions of this work are as follows.

  1. A general and systematic procedure for incorporating thermal fluctuations into central-moments-based lattice Boltzmann schemes is presented, in which stochastic forcing is introduced directly into the post-collision state in terms of central moments and consistently paired with mode-dependent relaxation.

  2. The formulation ensures exact equilibrium equipartition and compliance with the fluctuation–dissipation theorem for isothermal, weakly compressible flows, while preserving the defining properties of central-moments collision operators.

  3. Explicit fluctuating collision operators are derived for the D2Q9 and D3Q27 lattices, providing concrete and reproducible implementations in two and three spatial dimensions. The extension to reduced-velocity discretisation is also addressed.

  4. The schemes are assessed through numerical tests focusing on equilibrium velocity fluctuations, moment-wise energy partitioning, and robustness over a broad range of relaxation parameters. Extensions to multiphase, multi-component, or fully thermal formulations are not considered here and are left for future work.

Fluctuating lattice Boltzmann formulation

The construction of a fluctuating lattice Boltzmann method requires more than the formal addition of random noise to the standard collision operator. Thermal fluctuations must simultaneously (i) preserve exact conservation of mass and momentum, (ii) satisfy a discrete fluctuation–dissipation balance at the lattice level, and (iii) remain invariant under Galilean transformations. These requirements cannot, in general, be met directly in population space, where kinetic degrees of freedom are intrinsically coupled. The strategy adopted here is therefore to reformulate the collision step in a suitably chosen moment space, where conserved, hydrodynamic, and purely kinetic modes can be clearly separated and treated independently.

BGK fluctuating LBM

In absence of body forces, the discrete kinetic equation can be written as[4]

fi(x+ciΔt,t+Δt)=fi(x,t)+Ωi(x,t)+ζi(x,t),(1)f_i(\boldsymbol{x}+\boldsymbol{c}_i\Delta t,t+\Delta t)=f_i(\boldsymbol{x},t)+\Omega_i(\boldsymbol{x},t)+\zeta_i(\boldsymbol{x},t), \tag*{(1)}

where fi(x,t)f_i(\boldsymbol{x},t) denotes the particle distribution function associated with discrete velocity ci\boldsymbol{c}_i, with i=0,…,p−1i=0,\ldots,p-1, at spatial location x\boldsymbol{x} and time tt. The process is usually split into two sub-steps which are known as collision

fi⋆(x,t)=fi(x,t)+Ωi(x,t)+ζi(x,t),(2)f_i^{\star}(\boldsymbol{x},t)=f_i(\boldsymbol{x},t)+\Omega_i(\boldsymbol{x},t)+\zeta_i(\boldsymbol{x},t), \tag*{(2)}

and streaming

fi(x+ciΔt,t+Δt)=fi⋆(x,t),(3)f_i(\boldsymbol{x}+\boldsymbol{c}_i\Delta t,t+\Delta t)=f_i^{\star}(\boldsymbol{x},t), \tag*{(3)}

respectively, with Δt=1\Delta t=1. The superscript ⋆\star denotes post-collision quantities. To lighten the notation, let us drop the dependence on space and time in the rest of this section.

In its simplest form, the deterministic collision operator is expressed as a relaxation towards a local equilibrium[7], i.e.

Ωi=ω(fieq−fi),(4)\Omega_i=\omega\left(f_i^{\mathrm{eq}}-f_i\right), \tag*{(4)}

where the relaxation frequency ω\omega is related to the kinematic viscosity ν\nu through

ω=(νcs2+12)−1,(5)\omega=\left(\frac{\nu}{c_s^2}+\frac{1}{2}\right)^{-1}, \tag*{(5)}

and csc_s is the lattice sound speed. Macroscopic density and velocity fields are recovered as

ρ=∑ifi,ρu=∑icifi,(6)\rho=\sum_i f_i,\qquad\rho\boldsymbol{u}=\sum_i\boldsymbol{c}_i f_i, \tag*{(6)}

respectively. Thermal fluctuations are usually introduced through a stochastic term, namely ζi\zeta_i. Rather than injecting noise directly at the population level, fluctuations are generated in kinetic moment space and subsequently projected back onto the populations. This approach allows fluctuations to be applied selectively to non-conserved degrees of freedom while preserving the exact conservation laws. The stochastic contribution to the populations is written as

ζi=wi∑vbv−1eviϕvηv,(7)\zeta_i=w_i\sum_v b_v^{-1}e_{vi}\phi_v\eta_v, \tag*{(7)}

where ev=[ev0,…,ev,p−1]\boldsymbol{e}_v=[e_{v0},\ldots,e_{v,p-1}] denotes the basis vector associated with the vv-th kinetic mode, bv=∑iwievi2b_v=\sum_i w_i e_{vi}^2 is its normalisation factor, wiw_i is the lattice weight, and ηv\eta_v is an independent standard normal variable. The fluctuation amplitude ϕv\phi_v is determined by the discrete fluctuation–dissipation theorem. Let us underline that the index kk in Eq. (7) does not span the whole set of degrees of freedom (also known as modes or moments), but only the non-conserved ones[10].

Interpreting the stochastic collision operator as a Monte Carlo process satisfying detailed balance, the fluctuation amplitude takes the form

ϕv=ρkBTΛv(2−Λv)bvcs2,(8)\phi_v=\sqrt{\frac{\rho k_B T\Lambda_v(2-\Lambda_v)b_v}{c_s^2}}, \tag*{(8)}

where Λv\Lambda_v is the relaxation rate of the vv-th mode. This choice ensures exact balance between dissipation and fluctuations at the lattice level and recovers the Landau–Lifshitz fluctuating Navier–Stokes equations in the hydrodynamic limit.

Central-moments-based fluctuating LBM: Two-dimensional formulation

We consider a two-dimensional Cartesian system (x,y)(x,y) and the D2Q9 lattice (p=9p=9) with discrete velocities ci=[ci,x,ci,y]\boldsymbol{c}_i=[c_{i,x},c_{i,y}],

ci,x=[0,1,0,−1,0,1,−1,−1,1],c_{i,x}=[0,1,0,-1,0,1,-1,-1,1],
ci,y=[0,0,1,0,−1,1,1,−1,−1].(9)c_{i,y}=[0,0,1,0,-1,1,1,-1,-1]. \tag*{(9)}

The flow velocity vector is u=[ux,uy]\boldsymbol{u}=[u_x,u_y]. The equilibrium distribution fieqf_i^{\mathrm{eq}} is expanded via a Hermite expansion[19]

fieq=wi∑n=0N1n!cs2nHi,α0…αn(n)aα0,α1(n),(10)f_i^{\mathrm{eq}}=w_i\sum_{n=0}^{N}\frac{1}{n!c_s^{2n}}\mathcal{H}_{i,\alpha_0\ldots\alpha_n}^{(n)}a_{\alpha_0,\alpha_1}^{(n)}, \tag*{(10)}

where the corresponding lattice weights are

w0=49,w1−4=19,w5−8=136,(11)w_0=\frac{4}{9},\qquad w_{1-4}=\frac{1}{9},\qquad w_{5-8}=\frac{1}{36}, \tag*{(11)}

and the definition of the Hermite tensor up to the order N=4N=4 is

Hi(0)=1,\mathcal{H}_i^{(0)}=1,
Hi,α(1)=ci,α,\mathcal{H}_{i,\alpha}^{(1)}=c_{i,\alpha},
Hi,αβ(2)=ci,αβ−cs2δαβ,\mathcal{H}_{i,\alpha\beta}^{(2)}=c_{i,\alpha\beta}-c_s^2\delta_{\alpha\beta},
Hi,αβγ(3)=ci,αβγ−cs2(ci,αδβγ+ci,βδαγ+ci,γδαβ),\mathcal{H}_{i,\alpha\beta\gamma}^{(3)} =c_{i,\alpha\beta\gamma} -c_s^2(c_{i,\alpha}\delta_{\beta\gamma} +c_{i,\beta}\delta_{\alpha\gamma} +c_{i,\gamma}\delta_{\alpha\beta}),
Hi,αβγδ(4)=ci,αci,βci,γci,δ−cs2(ci,αci,βδγδ+ci,αci,γδβδ+ci,αci,δδβγ+ci,βci,γδαδ+ci,γci,δδαβ)+cs4(δαβδγδ+δαγδβδ+δαδδβγ),(12)\begin{aligned} \mathcal{H}_{i,\alpha\beta\gamma\delta}^{(4)} ={}&c_{i,\alpha}c_{i,\beta}c_{i,\gamma}c_{i,\delta} -c_s^2(c_{i,\alpha}c_{i,\beta}\delta_{\gamma\delta} +c_{i,\alpha}c_{i,\gamma}\delta_{\beta\delta}\\ &+c_{i,\alpha}c_{i,\delta}\delta_{\beta\gamma} +c_{i,\beta}c_{i,\gamma}\delta_{\alpha\delta} +c_{i,\gamma}c_{i,\delta}\delta_{\alpha\beta})\\ &+c_s^4(\delta_{\alpha\beta}\delta_{\gamma\delta} +\delta_{\alpha\gamma}\delta_{\beta\delta} +\delta_{\alpha\delta}\delta_{\beta\gamma}), \tag*{(12)} \end{aligned}

References

  1. [1]M. Polimeno, C. Kim, F. Blanchette, I. Srivastava, A. L. Garcia, A. J. Nonaka, and J. B. Bell, “Thermodynamic consistency and fluctuations in mesoscopic stochastic simulations of reactive gas mixtures,” J Chem Phys 162 (2025), 10.1063/5.0251770.
  2. [2]A. Donev, E. Vanden-Eijnden, A. Garcia, and J. Bell, “On the accuracy of finite-volume schemes for fluctuating hydrodynamics,” Commun Appl Math Comput Sci 5, 149–197 (2010).arxiv.org/abs/0906.2425
  3. [3]R. Adhikari, K. Stratford, M. E. Cates, and A. J. Wagner, “Fluctuating lattice Boltzmann,” Europhys Lett 71, 473 (2005).DOI
  4. [4]T. Krüger, H. Kusumaatmaja, A. Kuzmin, O. Shardt, G. Silva, and E. M. Viggen, The Lattice Boltzmann Method: Principles and Practice (Springer International Publishing, 2017).DOI
  5. [5]B. Dünweg, U. D. Schiller, and A. J. C. Ladd, “Statistical mechanics of the fluctuating lattice Boltzmann equation,” Phys Rev E 76, 036704 (2007).pubmed.ncbi.nlm.nih.gov/17930358
  6. [6]D. Belardinelli, M. Sbragaglia, L. Biferale, M. Gross, and F. Varnik, “Fluctuating multicomponent lattice Boltzmann model,” Phys Rev E 91, 023313 (2015).pubmed.ncbi.nlm.nih.gov/25768641
  7. [7]P. L. Bhatnagar, E. P. Gross, and M. Krook, “A model for collision processes in gases. i. small amplitude processes in charged and neutral one-component systems,” Phys Rev 94, 511–525 (1954).
  8. [8]D. d’Humieres, “Multiple-relaxation-time lattice Boltzmann models in three dimensions,” Philos T R Soc A 360, 437–451 (2002).
  9. [9]A. J. Wagner and K. Strand, “Fluctuating lattice Boltzmann method for the diffusion equation,” Phys Rev E 94, 033302 (2016).pubmed.ncbi.nlm.nih.gov/27739715
  10. [10]M. Lauricella, A. Montessori, A. Tiribocchi, and S. Succi, “Regularized fluctuating lattice Boltzmann model,” J Chem Phys 163, 161402 (2025).DOI
  11. [11]M. Lauricella, A. Montessori, A. Tiribocchi, and S. Succi, “Ghost-mode filtered fluctuating lattice Boltzmann method,” arXiv preprint arXiv:2601.03806 (2026), 10.48550/arXiv.2601.03806.arxiv.org/abs/2601.03806
  12. [12]G. A. Greiner, and J. Korvink, “Cascaded digital lattice Boltzmann automata for high Reynolds number flow,” Phys Rev E 73, 066705 (2006).
  13. [13]P. Asinari, “Generalized local equilibrium in the cascaded lattice Boltzmann method,” Phys Rev E 78, 016701 (2008).pubmed.ncbi.nlm.nih.gov/18764075
  14. [14]X. Shan, “Central-moment-based Galilean-invariant multiple-relaxation-time collision model,” Phys Rev E 100, 043308 (2019).DOI
  15. [15]C. Coreixas, G. Wissocq, B. Chopard, and J. Latt, “Impact of collision models on the physical properties and the stability of lattice Boltzmann methods,” Phil Trans R Soc A 378 (2020), 10.1098/rsta.2019.0397.DOI
  16. [16]L. Fei, K. H. Luo, and Q. Li, “Three-dimensional cascaded lattice Boltzmann method: Improved implementation and consistent forcing scheme,” Phys Rev E 97, 053309 (2018).DOI
  17. [17]S. Saito, A. De Rosis, A. Festuccia, K. Kaneko, Y. Abe, and K. Koyama, “Color-gradient lattice Boltzmann model with nonorthogonal central moments: Hydrodynamic melt-jet breakup simulations,” Phys Rev E 98, 013305 (2018).DOI
  18. [18]Y. Zhou, M. C. Santasmasas, A. De Rosis, I. Hinder, C. Moulinet, and A. Revell, “A coupled finite volume-lattice Boltzmann method for incompressible internal flows,” Comput Phys Commun 314, 109686 (2025).DOI
  19. [19]C. Coreixas, G. Wissocq, G. Puigt, J.-F. m. c. Boussuges, and P. Sagaut, “Recursive regularization step for high-order lattice Boltzmann methods,” Phys Rev E 96, 033306 (2017).DOI
  20. [20]Y. Ning, K. N. Premnath, and D. V. Patil, “Numerical study of the properties of the central moment lattice Boltzmann method,” Int J Numer Meth Fl 82, 59–90 (2016).arxiv.org/abs/1202.6351
  21. [21]L. Fei, J. Du, K. H. Luo, S. Succi, M. Lauricella, A. Montessori, and Q. Wang, “Modeling realistic multiphase flows using a non-orthogonal multiple-relaxation-time lattice Boltzmann method,” Phys Fluids 31, 042105 (2019).DOI
  22. [22]A. De Rosis, “Non-orthogonal central moments relaxing to a discrete equilibrium: A D2Q9 lattice Boltzmann model,” Europhys Lett 116, 44003 (2016).
  23. [23]O. Malaspinas, “Increasing stability and accuracy of the lattice Boltzmann scheme: recursivity and regularization,” (2015), arXiv:1505.06900 [physics.flu-dyn].arxiv.org/abs/1505.06900
  24. [24]A. De Rosis, “Nonorthogonal central-moments-based lattice Boltzmann scheme in three dimensions,” Phys Rev E 95, 013310 (2017).
  25. [25]A. De Rosis, R. Huang, and C. Coreixas, “Universal formulation of central-moments-based lattice Boltzmann method with external forcing for the simulation of multiphysics phenomena,” Phys Fluids 31, 117102 (2019).
  26. [26]A. De Rosis and K. H. Luo, “Role of higher-order hermite polynomials in the central-moments-based lattice Boltzmann framework,” Phys Rev E 99, 013301 (2019).
  27. [27]A. De Rosis and C. Coreixas, “Multiphysics flow simulations using D3Q19 lattice Boltzmann methods based on central moments,” Phys Fluids 32, 117101 (2020).
  28. [28]C. R. Trott, D. Lebrun-Grandie, D. Arndt, J. Ciesko, V. Dang, N. Ellingwood, R. Gayatri, E. Harvey, D. S. Hollman, D. Ibanez, N. Jilek, J. Madsen, J. Miles, D. Poliakoff, A. Powell, S. Rajamanickam, M. Simberg, D. Sunderland, B. Turcksin, and J. Wilke, “Kokkos 3: Programming model extensions for the exascale era,” IEEE Trans Parallel Distrib Syst 33, 805–817 (2022).

Paper details

vlnc:W_uw56bj35
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

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