Skip to content
Licensed Unlicensed Requires Authentication Published by De Gruyter September 29, 2020

Convergence of time-splitting approximations for degenerate convection–diffusion equations with a random source

  • Roberto Díaz-Adame and Silvia Jerez EMAIL logo

Abstract

In this paper we propose a time-splitting method for degenerate convection–diffusion equations perturbed stochastically by white noise. This work generalizes previous results on splitting operator techniques for stochastic hyperbolic conservation laws for the degenerate parabolic case. The convergence in Llocp of the time-splitting operator scheme to the unique weak entropy solution is proven. Moreover, we analyze the performance of the splitting approximation by computing its convergence rate and showing numerical simulations for some benchmark examples, including a fluid flow application in porous media.

JEL Classification: 65M06; 65M12; 65Z05

Funding statement: This work was partially supported by Mexico CONACyT Project CB2016-286437.

A Numerical methods

A.1 Switch flux-limiter scheme

To solve numerically the deterministic parabolic equation (1.5) with an initial condition u(x, 0) = u0(x), for x ∈ [a, b] ∈ ℝ, we consider a uniform grid for the spatial-time domain. Denoting by (xj, tn) the mesh point with tn = n(Δt) and xj = a + jΔx, n, j ∈ ℕ, being a mesh cell Cjn = (xj−1/2, xj+1/2) × (tn−1/2, tn+1/2). The Switch flux-limiter approximation [10] for problem (1.5) is

Ujn+1=UjnΔtΔxHj+1/2Hj1/2

where

Hj+1/2=12φj+1/2nFj+1/2UP+12(2φj+1/2n)[Fj+1/2RI+aj+1/2φj+1/212×(Uj+1nUjnΔtΔx[(λj+f(Uj+1n)λjf(Uj+1n))(λj+f(Ujn)λj+f(Ujn))])]K(Uj+1n)K(Ujn)Δx.(A.1)

with aj+1/2=f12(Ujn+Uj+1n)Δt2Δx(f(Uj+1n)f(Ujn)),λj+=max{f(Ujn)/|f(Ujn)|,0},andλj=min{f(Ujn)/|f(Ujn)|,0}. A flux-limiter scheme is a hybrid method based on a combination of two standard schemes, in the Switch case: FRI denotes the Richtmyer two-step Lax Wendroff method and FUP denotes a conservative upwind method known as Steger–Warming scheme. These numerical fluxes are defined as

Fj+1/2RI=f(Ujn+Uj+1n2)Δt2Δx(f(Uj+1n)f(Ujn))Fj+1/2UP=λj+f(Ujn)λjf(Uj+1n).

The associated weight function is

φj+1/2:=121πarctan|ϑjn|1

by measuring the smoothness of the advective flux function with

ϑjn=(UjnUj1n)/(Uj+1nUjn),f(Ujn)0(Uj+1nUjn)/(UjnUj1n),f(Ujn)<0.

Finally, an important term of this scheme is the switch function

φj+1/2n=0,|ϑjn1|<1/4and(Uj+1nUjn),(UjnUj1n)>02,otherwise

which turns on and off the upwind flux, modifying the algorithm in presence of discontinuous or smooth regions of the solution. In [10], authors proved that the viscous Switch flux-limiter scheme (A.1) for equation (1.5) is TVD-stable and convergent if

11|ajn|maxj,n{|ajn|/djn}

with |ajn| < 1 for convex fluxes and 1/2 ⩽ |ajn| ⩽ 1 for non convex fluxes, where ajn = (Δt/Δx)(f /u)|j,n and djn=((Δx)2)/Δt)d(ujn).

A.2 Linear Steklov method

Such iterative method was proposed in [11] for solving the stochastic differential equation (1.6). Discretizing the time domain [0, T] with a constant time-step Δt such that tn = nΔt for some n ∈ ℕ. Denoting by Un the approximation and Wn the discrete standard Brownian motion satisfying ΔWn:=(Wtn+1Wtn)=ΔtVn with VnN(0, 1). Using the integral formulation of equation (1.6), we can rewrite the solution process at time tn+1 as follows:

u(tn+1)=u(tn)+tntn+1g(u(s))ds+tntn+1h(u(s))dW(s),tnttn+1.

The Itô integral is approximated by

tntn+1h(u(s))dW(s)h(Un)ΔWn.

Assuming that the drift function g(u) can be linearized of the following form

g(u)=a1(u)u

where a1 is a scalar function, then the linear Steklov iterative process is written as

Un+1=UneΔta1(Un)+h(Un)ΔWn.(A.2)

With regular conditions on local Lipschitz functions g and h satisfying a monotony property, the linear Steklov method becomes a robust and strongly convergent algorithm with a standard order of one-half.

References

[1] C. Bauzet, G. Vallet, and P. Wittbold, A degenerate parabolic–hyperbolic Cauchy problem with a stochastic force, J. Hyperbol. Differ. Equ., 12 (2015), 501–533.10.1142/S0219891615500150Search in Google Scholar

[2] C. Bauzet, Time-splitting approximation of the Cauchy problem for a stochastic conservation law, Math. Comput. Simulat., 118 (2015), 73–86.10.1016/j.matcom.2014.11.012Search in Google Scholar

[3] R. Bürger, R. Ruiz, K. Schneider, and M. Sepúlveda, Fully adaptive multiresolution schemes for strongly degenerate parabolic equations in one space dimension, ESAIM-Math. Model. Num., 42 (2008), 535–563.10.1051/m2an:2008016Search in Google Scholar

[4] R. J. DiPerna, Measure-valued solutions to conservation laws, Arch. Ration. Mech. Anal., 88 (1985), 223–270.10.1007/BF00752112Search in Google Scholar

[5] S. Evje and K. H. Karslen, Monotone Difference Approximations of BV Solutions to Degenerate Convection–Diffusion Equations, SIAM J. Numer. Anal., 37 (2000), 1838–1860.10.1137/S0036142998336138Search in Google Scholar

[6] S. Evje, K. H. Karlsen, K. A. Lie, and N. H. Risebro, Front tracking and operator splitting for nonlinear degenerate convection–diffusion equations. In: Parallel Solution of Partial Differential Equations, Springer, New York, 2000, pp. 209–227.10.1007/978-1-4612-1176-1_9Search in Google Scholar

[7] R. Eymard, T. Gallouët, and R. Herbin, Existence and uniqueness of the entropy solution to a nonlinear hyperbolic equation, Chinese Ann. Math. Ser. B, 16 (1995), 1–14.Search in Google Scholar

[8] H. Holden and N. H. Risebro, Conservation laws with a random source, Appl. Math. Optim., 36 (1997), 229–241.10.1007/s002459900061Search in Google Scholar

[9] M. Hutzenthaler, A. Jentzen, and P. E. Kloeden, Strong and weak divergence in finite time of Euler’s method for stochastic differential equations with non-globally Lipschitz continuous coefficients, Proc. Royal Soc. London Ser. A, 467 (2011), 1563–1576.10.1098/rspa.2010.0348Search in Google Scholar

[10] S. Jerez and M. Arciga, Switch flux limiter method for viscous and nonviscous conservation laws, Appl. Math. Comput., 246 (2014), 292–305.10.1016/j.amc.2014.08.011Search in Google Scholar

[11] S. Jerez and S. Díaz-Infante, Exact difference schemes for stochastic differential equations. In: Exact Finite-Difference Schemes, De Gruyter, 2016, pp. 204–218.10.1515/9783110491326-008Search in Google Scholar

[12] S. Jerez and C. Parés, Entropy stable schemes for degenerate convection–diffusion equations, SIAM J. Numer. Anal., 55 (2017), 240–264.10.1137/16M1076411Search in Google Scholar

[13] K. H. Karlsen and N. H. Risebro, On the uniqueness and stability of entropy solutions of nonlinear degenerate parabolic equations with rough coefficients, Discrete Contin. Dyn. Syst. Ser. A, 9 (2003), 1081–1104.10.3934/dcds.2003.9.1081Search in Google Scholar

[14] J. D. R. Kersner, On a nonlinear degenerate parabolic equation in infiltration or evaporation through a porous medium, J. Differ. Equations, 69 (1987), 368–403.10.1016/0022-0396(87)90125-2Search in Google Scholar

[15] Y. Liu, C. W. Shu, and M. Zhang, High order finite difference WENO schemes for nonlinear degenerate parabolic equations, SIAM J. Sci. Comput., 33 (2011), 939–965.10.1137/100791002Search in Google Scholar

[16] J. D. Logan, Transport Modeling in Hydrogeochemical Systems, Springer, New York, 2001.10.1007/978-1-4757-3518-5Search in Google Scholar

[17] A. Michel and J. Vovelle, Entropy formulation for parabolic degenerate equations with general Dirichlet boundary conditions and application to the convergence of FV methods, SIAM J. Numer. Anal., 41 (2003), 2262–2293.10.1137/S0036142902406612Search in Google Scholar

[18] F. Peyroutet, Etude d’une méthode Splitting pour des lois de conservation scalaires avec terme de source, Ph.D. thesis, U.P.P.A., 1999.Search in Google Scholar

[19] J. Proft and B. Riviere, Discontinuous Galerkin methods for convection–diffusion equations for varying and vanishing diffusivity, Int J. Numer. Anal. Mod., 6 (2009), 533–561.Search in Google Scholar

[20] G. Vallet, Uniqueness of the measure-valued solution to a conservation law with boundary conditions, Appl. Math. Lett., 11 (1998), 91–94.10.1016/S0893-9659(98)00062-7Search in Google Scholar

[21] G. Vallet, Stochastic perturbation of nonlinear degenerate parabolic problems, Differ. Integral Equ., 21 (2008), 1055–1082.Search in Google Scholar

Received: 2020-01-30
Revised: 2020-08-28
Accepted: 2020-09-13
Published Online: 2020-09-29
Published in Print: 2021-03-26

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 28.4.2024 from https://www.degruyter.com/document/doi/10.1515/jnma-2020-0012/html
Scroll to top button