Numerical approximations to statistical conservation laws for scalar hyperbolic equations
Qian Huang*, Christian Rohde · 2025
Read Numerical approximations to statistical conservation laws for scalar hyperbolic equations by Qian Huang*, Christian Rohde (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-05039.txt.
Introduction
Turbulence can be described through the chance spreads of the flow, but the equations for those spreads contain unclosed terms. Qian Huang and one co-author study approximations to such conservation laws in a 2025 applied mathematics paper. Their setting is scalar hyperbolic balance laws, a simpler model for turbulent transport.
They show those terms stay important even as viscosity vanishes, and they keep probability densities from going negative. A sampling-based estimator, built from runs of the underlying solutions, approximates those terms. Error bounds link guess quality to estimation error. Tests on solvable problems confirm the guess converges as samples increase.
Catalog record
| CREATOR | Qian Huang*, Christian Rohde |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-05039
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Numerical approximations to statistical conservation laws for scalar hyperbolic equations Thanks: * Corresponding author Qian Huang*, Christian Rohde Abstract . Motivated by the statistical description of turbulence, we study statistical conservation laws in the form of kinetic-type PDEs for joint probability density functions (PDFs) and cumulative distribution functions (CDFs) associated with solutions of scalar balance laws. Starting from viscous balance laws, the resulting PDF/CDF equations involve unclosed conditional averages arising in the viscous terms. We show that these terms exhibit a dissipative anomaly: they remain non-negligible in the vanishing viscosity limit and are essential to preserve the nonnegativity of evolving PDFs. To approximate these PDF/CDF equations in a unified framework, we propose a novel sampling-based estimator for the unclosed terms, constructed from numerical or exact realizations of the underlying balance-law solutions. In certain cases, a priori error bounds can be derived, demonstrating that the deviation between the true and approximate CDFs is controlled by the estimation error of the unclosed terms. Numerical experiments with analytically solvable test problems confirm that the sampling-based approximation converges satisfactorily with the number of samples. 2020 Mathematics Subject Classification Primary 35L65 · 35R60 · 35Q35 1. Introduction The study of statistical conservation laws is motivated by the inherently statistical nature of turbulence, where uncertain measurements lead to markedly different predictions [ 26 ] . The fluid velocity and pressure (u_{f},p_{f}):(t,x)\in\mathbb{R}_{+}\times\mathbb{R}^{d}\mapsto\mathbb{R}^{d}\times\mathbb{R} are governed by the incompressible Navier-Stokes equations (NSE), and uncertainties are naturally modeled as random perturbations of the initial and/or boundary data. Statistical conservation laws refer to kinetic equations (conservation laws) for the joint probability density functions (PDFs) f^{(K)}=f^{(K)}(t,x_{1},v_{1},\dots,x_{K},v_{K}) of the velocity field u_{f} at multiple spatial locations \{x_{i}\}_{1}^{K} ( K\in\mathbb{N} ) and time instants [ 28 , 16 ] . For Q^{i}\subset\mathbb{R}^{d} Borel sets ( i=1,\dots,K ), we have (1.1) \text{Prob}(\{u_{f}(t,x_{i})\in Q^{i}\text{ for all }i\})=\int_{\prod Q^{i}}f^{(K)}(t,x_{1},v_{1},\dots,x_{K},v_{K})\,dv_{1}\cdots dv_{K}. These joint PDFs encode the complete statistical information of the turbulent velocity field. For example, the mean velocity \langle u_{f}\rangle(t,x) and the two-point correlation S_{r}(\ell;x)=\langle|(u_{f}(x+\ell n)-u_{f}(x))\cdot n|^{r}\rangle (with n\in\mathbb{R}^{d} a unit vector and \ell>0 ) are moments of f:=f^{(1)}(t,x,v) and f^{(2)}(t,x,v_{1},x+\ell n,v_{2}) , respectively. By \langle\cdot\rangle we denote the mean with respect to the given probability measure. The Reynolds-averaged NSE governing \langle u_{f}\rangle are central in engineering practice [ 32 ] . For near-wall shear flows, \langle u_{f}\rangle obeys a logarithmic law [ 32 ] . As for the structure function S_{r}(\ell;x) , it follows Kolmogorov’s scaling law S_{r}(\ell;x)\sim\ell^{\zeta_{r}} with \zeta_{r}=r/3 in the inertial range for d=3 [ 22 ] . This scaling holds for r=2,3 , but shows significant deviations for larger r [ 21 ] . Due to the limited spatio-temporal validity of these phenomenological theories, statistical conservation laws provide a promising unified framework for capturing velocity statistics across all scales [ 16 ] . Generally they read as \partial_{t}f^{(K)}=L_{K}f^{(K)} , with L_{K} a problem-specific operator. As with many statistical formulations of NSE, however, these equations are not closed: \partial_{t}f^{(K)}=C_{K}f^{(K)}+U_{K} , where C_{K} is known and U_{K} the unclosed part, typically arising from pressure and viscous terms in NSE. This term can be expressed as L_{K,K+1}(f^{(K)},f^{(K+1)}) , forming the infinite Lundgren-Monin-Novikov (LMN) hierarchy [ 24 , 25 , 27 ] , or equivalently through unknown conditional averages such as \langle\Delta_{x_{i}}u_{f}|u_{f}=v_{i}\rangle [ 16 ] . Various analytical closures for U_{K} have been proposed, including statistical independence closures [ 35 ] , ‘Markov-process-in-scale’ ansatz for velocity increments [ 17 , 15 ] , and DNS-data-driven closures [ 6 ] . Statistical conservations laws also arise in other incompressible flow problems. The ‘PDF method’ for turbulent combustion employs velocity-composition joint PDFs to model turbulence-reaction coupling in the incompressible NSEs with varying densities [ 31 ] . Passive scalar transport in random (incompressible) velocity fields provides another classical example, with long-time scalar PDF behavior analyzed in [ 33 ] . By contrast, statistical conservation laws for compressible flows and systems of hyperbolic balance laws remain largely unexplored. When the divergence-free condition is absent, even the spatial transport term of the kinetic equations acquires nonlocal contributions (see Appendix A for preliminary PDF equations for 1D systems). These PDF equations may connect to the theory of statistical solutions of conservation laws [ 13 , 14 , 23 , 18 ] . A key difficulty is dimensionality: the K -point PDF in ( 1.1 ) depends on 1+2dK variables. Therefore, most existing work focuses on the scalar setting: (1.2) \begin{split}&\partial_{t}u+\nabla_{x}\cdot g(u)=Q(u,\nabla_{x}u,\Delta_{x}u),\\ &u(0,x,\xi(\omega))=u_{0}(x,\xi(\omega)),\end{split}\quad t>0,x\in D\subset\mathbb{R}^{d},\omega\in\Omega, where u=u(t,x,\xi(\omega))\in\mathbb{R} , g=g(u) is a smooth flux in \mathbb{R}^{d} , \xi:\Omega\to\mathbb{R}^{N_{\xi}} is an absolutely-continuous random variable with a prescribed density function p(\xi) , and Q represents possible source terms. For scalar conservation laws ( Q=0 ), the ‘method of distributions’ derives governing equations for single-point cumulative distribution functions (CDFs), which involve only spatial transport [ 34 , 7 ] . If shocks occur in the solution u , a kinetic defect source term must be added [ 29 ] , which is generally unknown a prior but may be inferred from data [ 7 ] . When Q=Q(u) , the PDF equation is closed for smooth u [ 36 ] , equivalent to the mentioned CDF formulation (see also Section 2 ). For Q=Q(u,\nabla_{x}u) , closed equations for the joint u – \nabla_{x}u PDF can be derived in one dimension for smooth solutions [ 36 ] , while discontinuities will again introduce defect terms to these kinetic equations. Another important line of research concerns the viscous Burgers equation with random forcing, corresponding to g(u)=u^{2}/2 and Q=\epsilon\Delta_{x}u in ( 1.2 ), often viewed as a scalar analogue of NSE without pressure [ 2 , 4 ] . Governing equations for single- and two-point PDFs have been analyzed, some with analytical closure proposals [ 30 , 19 , 11 ] . Notably, these studies clarified the statistics of velocity gradients and resolved a long-standing debate about their tail behavior [ 11 ] . The objective of this paper is to investigate numerical approximations to statistical conservation laws for hyperbolic balance laws, focusing on the scalar case ( 1.2 ). Unlike earlier studies [ 36 , 9 , 7 ] , we consider viscous balance laws with Q=\epsilon\Delta_{x}u in ( 1.2 ) to mimic the interplay between hyperbolic transport and viscosity operators. The viscous term ensures global existence of classical solutions of u , but leads once again to unclosed statistical conservation laws. In Section 2 we systematically derive these laws. A previously-undocumented dissipative anomaly is further revealed: for nonlinear fluxes g(u) in ( 1.2 ), the viscosity-related unclosed term U_{K} remains essential even in the vanishing viscosity limit, as it guarantees the positivity of PDFs. Since analytical closures of U_{K} often introduce very different model structures, our approach does not attempt to derive them. Instead, we approximate U_{K} directly from sampled solutions of u , a simple yet, to our knowledge, unexplored pipeline. This sampling-based method is flexible and extends naturally to statistical conservation laws of other systems, including NSE. To validate it, we construct analytical test cases for ( 1.2 ). Numerical results confirm the a priori error analysis and show that PDF/CDF errors diminish with increasing sample size. Finally, we remark that the present setting ( 1.2 ) coincides with random hyperbolic conservation laws. Many numerical strategies