\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
Citation: Abdulkarim Hassan Ibrahim, Poom Kumam, Auwal Bala Abubakar, Umar Batsari Yusuf, Seifu Endris Yimer, Kazeem Olalekan Aremu. An efficient gradient-free projection algorithm for constrained nonlinear equations and image restoration[J]. AIMS Mathematics, 2021, 6(1): 235-260. doi: 10.3934/math.2021016
[1] | Hongying Jiao, Shuhai Zhu, Jinguo Zhang . Existence of infinitely many solutions for critical sub-elliptic systems via genus theory. Communications in Analysis and Mechanics, 2024, 16(2): 237-261. doi: 10.3934/cam.2024011 |
[2] | Jizheng Huang, Shuangshuang Ying . Hardy-Sobolev spaces of higher order associated to Hermite operator. Communications in Analysis and Mechanics, 2024, 16(4): 858-871. doi: 10.3934/cam.2024037 |
[3] | Xiulan Wu, Yaxin Zhao, Xiaoxin Yang . On a singular parabolic p-Laplacian equation with logarithmic nonlinearity. Communications in Analysis and Mechanics, 2024, 16(3): 528-553. doi: 10.3934/cam.2024025 |
[4] | Erlend Grong, Irina Markina . Harmonic maps into sub-Riemannian Lie groups. Communications in Analysis and Mechanics, 2023, 15(3): 515-532. doi: 10.3934/cam.2023025 |
[5] | Leandro Tavares . Solutions for a class of problems driven by an anisotropic (p,q)-Laplacian type operator. Communications in Analysis and Mechanics, 2023, 15(3): 533-550. doi: 10.3934/cam.2023026 |
[6] | Velimir Jurdjevic . Time optimal problems on Lie groups and applications to quantum control. Communications in Analysis and Mechanics, 2024, 16(2): 345-387. doi: 10.3934/cam.2024017 |
[7] | Zhiyong Wang, Kai Zhao, Pengtao Li, Yu Liu . Boundedness of square functions related with fractional Schrödinger semigroups on stratified Lie groups. Communications in Analysis and Mechanics, 2023, 15(3): 410-435. doi: 10.3934/cam.2023020 |
[8] | Ming Liu, Binhua Feng . Grand weighted variable Herz-Morrey spaces estimate for some operators. Communications in Analysis and Mechanics, 2025, 17(1): 290-316. doi: 10.3934/cam.2025012 |
[9] | Xiaotian Hao, Lingzhong Zeng . Eigenvalues of the bi-Xin-Laplacian on complete Riemannian manifolds. Communications in Analysis and Mechanics, 2023, 15(2): 162-176. doi: 10.3934/cam.2023009 |
[10] | Eleonora Amoroso, Angela Sciammetta, Patrick Winkert . Anisotropic (→p,→q)-Laplacian problems with superlinear nonlinearities. Communications in Analysis and Mechanics, 2024, 16(1): 1-23. doi: 10.3934/cam.2024001 |
In this paper, we focus on the following hydrodynamically consistent Cahn-Hilliard phase field model for incompressible two-phase flows with variable density
ρt+∇⋅(ρu)=0, | (1.1a) |
ϕt+u⋅∇ϕ−γΔw=0, | (1.1b) |
w=−Δϕ+1ε2(ϕ3−ϕ), | (1.1c) |
ρ(ut+u⋅∇u)−ηΔu+∇p−λw∇ϕ=0, | (1.1d) |
∇⋅u=0, | (1.1e) |
where ϕ, w, u and p are the phase field function, the chemical potential, the velocity of flow and pressure respectively. Moreover, ρ=12(ρ1+ρ2)+ϕ2(ρ1−ρ2) is the density, where ρ1 and ρ2 are the densities of the two fluids. The parameters γ, ε, η and λ are the mobility parameter related to the relaxation time scale, the interface thickness [1], the viscosity of the field, and the mixing energy density respectively. The system (1.1) is supplemented with appropriate boundary and initial conditions which are given as
ϕ(x,0)=ϕ0(x), ρ(x,0)=ρ0(x), u(x,0)=u0(x),ϕ|∂Ω=0, u|∂Ω=0. |
This model is also called the Cahn-Hilliard-Navier-Stokes model when the density is constant, which has many practical applications in physical and engineering, such as wetting, coating, and painting. By considering the influence of variable density, this model has broad applicability, which include highly stratified flows, interfaces between fluids of different densities and some problems of inertial confinement.
The phase field method which has a wide range of applications is one of the main methods to deal with the fluid interface in two-phase flow modeling; see [2,3,4,5] and the references therein. It was initially developed to simulate solid-liquid phase transitions, where the interface is treated as a thin, smooth transition layer[6,7,8] to remove the singularities between two phases. The basic framework is to use phase field variables to represent the volume fraction of fluid components and then adopt a variational form to derive the model. Recently, it has increasingly attracted researchers' interest, mainly because the phase field method is superior to other available methods in some aspects of two-phase flow [9]. Various material properties or complex interface behaviors can be simulated directly by introducing suitable energy functions. Numerical solutions play a crucial role in their study and applications because the analytic solutions are usually not available.
A few works have been devoted to the design, analysis and implementation of numerical schemes for the phase field model, although this model is very classical and canonical. We briefly review the available methods used in the phase field model. It is worth noting that there are projection/gauge/penalty methods [10,11,12], scalar auxiliary variables [13,14], linear stability [15,16], convex splitting [17,18,19], invariable energy quadratization(IEQ) [20,21], nonlinear quadratic [22], exponential time differencing [23], etc. In practical applications, we often couple the flow-field equation with the phase field equation. Typically two-phase incompressible flow models are coupled to phase-field models. The Cahn-Hilliard model is taken into account in this paper, because it is effective in the following two aspects: (ⅰ) Cahn-Hilliard models can accurately conserve the volume and dynamics; (ⅱ) the equation is one of the most important models in mathematical physics. Because of these reasons, here we use the Cahn-Hilliard equation developed in [24] to couple the incompressible flows with variable density.
There also have been a lot of works on numerical approximates for the Cahn-Hilliard phase field model for incompressible two-phase flows with variable density. Hohenberg and Halperin proposed the model in [25] to simulate two incompressible viscous fluids with constant density. In [26], Gurtin et al. obtained the equal model by using the framework of rational continuum mechanics. A fully adaptive energy stabilization scheme is proposed in [27]. An efficient Picard iteration procedure was designed in [28] to further decouple the model. In the last years, many authors have been concerned with designing incompressible two-phase flow models with variable densities. Several efficient and energy-stable time discretization schemes for the coupled nonlinear Cahn-Hilliard phase field system with variable density are constructed; see [29]. Yang and Dong presented an energy-stable scheme in [30] for the numerical approximation of the two-phase governing equations with variable density and viscosity for the two fluids by introducing a scalar-valued variable related to the total of the kinetic energy and the potential free energy. A second-order accurate, coupled, energy-stable schemeis proposed in [31], where the Crank-Nicolson method and the IEQ method were used. In [32], the conservation scheme of the first-order energy law was established, in which the Cahn-Hilliard solver was used to decouple from the two-phase incompressible flows solver through the use of the fractional step method. Ye et al.[33] have designed a fully-decoupled type scheme to solve the Cahn-Hilliard phase field model for a two-phase incompressible fluid flow system with constant density. They only give a detailed practical implementation method and also prove the solvability. None of the various existing schemes have been subjected to error analysis, where the main difficulty lies in the delicate treatment of a several of nonlinear terms. Rigorous error estimates of models with variable densities, using an optimal order error bound, may seem to be a difficult prospect, but is a very interesting direction for future research.
In this paper, we finally arrive at an unconditionally stable in energy, first-order time-accurate scheme for the incompressible Cahn-Hilliard two phase flows with variable density by coming up with a fractional step method. This method has an advantage over the projection method in that the original boundary conditions of the problem can be implemented in all substeps of the scheme. The popular approach to discretizing the Cahn-Hilliard phase field model (1.1b) and (1.1c) in time is based on the convex-splitting of the free energy functional, i.e., an idea that can be traced back to [32]. In the convex-splitting framework, one treats the contribution from the convex part implicitly and the contribution from the concave part explicitly. This treatment promotes the energy stability of the scheme and this property is unconditional in terms of time steps. We also give a rigorous proof of the convergence results and error estimates in the theoretical analysis. The main contribution of this paper is a rigorous error analysis, particularly under the condition that energy stability is available. To the best of the authors' knowledge, the proof developed in this article is the first to have the description of error estimation. The accuracy and stability are also demonstrated through the simulation of various numerical examples, where the challenge is in creating an efficient and easy to implement numerical scheme that preserves the energy dissipation law.
The rest of this article is organized as follows. In Section 2, we construct an efficient time discrete scheme for variable density and derive unconditional energy stability. In Section 3, the error analysis of the semi-discrete scheme in time is provided. Some numerical experimentations are given in Section 4. Finally, conclusions are drawn in Section 5.
For the sake of simplicity, some notations are needed for the following content. We assume that the domain Ω∈R2 is open, sufficiently smooth and bounded. For any two functions ϕ(x) and ψ(x), their L2 inner product on Ω is denoted by (ϕ,ψ)=∫Ωϕ(x)ψ(x)dx, and the L2 norm of ϕ(x) is denoted by ‖ϕ‖=(ϕ,ϕ)12. Let τ>0 be the time step size and set tn=nτ for 0≤n≤N with T=Nτ. Moreover, we introduce the following spaces,
H={u∈L2(Ω)|divu in Ω, u⋅n=0 on ∂Ω},V=H10, V0={u∈V|divu=0 in Ω},M=L20(Ω)={q∈L2(Ω)|∫Ωqdx=0}, |
where n is the outward normal vector of ∂Ω. Next, we reformulate the system (1.1) as follows:
ρn+1−ρnτ+∇ρn+1⋅un=0, | (2.1a) |
ϕn+1−ϕnτ+˜un+1⋅∇ϕn−γΔwn+1=0, | (2.1b) |
wn+1=1ε2((ϕn+1)3−ϕn)−Δϕn+1, | (2.1c) |
ρn˜un+1−unτ−ηΔ˜un+1+ρn+1(un⋅∇)˜un+1−λwn+1∇ϕn+14ρn+1(∇⋅un)˜un+1=0, | (2.1d) |
ρn+1un+1−˜un+1τ−η(Δun+1−Δ˜un+1)+∇pn+1=0, | (2.1e) |
∇⋅un+1=0 | (2.1f) |
Remark 1. In (2.1d), the term 14ρn+1(∇⋅un)˜un+1 is added to obtain the unconditional stability, as it is 0 if ∇⋅un=0.
Theorem 1. (Stabilityofρ) For any τ>0 and any sequence {un}n=0,…,N satisfying the boundary condition un⋅n=0 on ∂Ω, the solution {ρn}n=1,…,N to (2.1a) satisfies
‖ρN‖2+N−1∑n=1‖ρn+1−ρN‖2=‖ρ0‖2. | (2.2) |
Proof. Testing (2.1a) with 2τρn+1 gives
‖ρn+1‖2−‖ρn‖2+‖ρn+1−ρn‖2+2τ∫Ω(un⋅∇ρn+1+12ρn+1∇⋅un)ρn+1dx=0. | (2.3) |
Owing to the boundary conditions on un, we note that
∫Ω(un⋅∇ρn+1+12ρn+1∇⋅un)ρn+1dx=12∫Ω∇⋅(|ρn+1|2un)dx=12∫∂Ω|ρn+1|2un⋅ndx=0. | (2.4) |
Thus, we get
‖ρn+1‖2+‖ρn+1−ρn‖2=‖ρn‖2. |
Summing all indices n ranging 0 to N−1, the proof is completed.
Next, the stability of (2.1b)–(2.1e) will be proved in the theorem below. Moreover, since the kinetic energy of the fluid is 12‖√ρnun‖2, it is more suitable to establish a bound based on ‖√ρnun‖2 than on the velocity itself. For simplicity, let us say that σn=√ρn for all 1≤n≤N and σ0=√ρ0.
Theorem 2. (Stabilityofenergy) For any τ>0, (2.1b)–(2.1e) satisfy the the conditions of following energy estimates:
‖σNuN‖2+λ‖∇ϕN‖2−λε2‖ϕN‖2+λ2ε2‖ϕN‖4L4+2τγλN−1∑n=0‖∇wn+1‖2+λN−1∑n=0(‖∇(ϕn+1−ϕn)‖2+12ε2‖ϕn+1−ϕn‖2)+λ2ε2N−1∑n=0(‖ϕn+1‖2−‖ϕn‖)2+N−1∑n=0(‖σn(˜un+1−un)‖2+‖σn+1(un+1−˜un+1)‖2)+ητN−1∑n=0(‖∇un+1‖2+‖∇˜un+1‖2+‖∇(un+1−˜un+1)‖2)+λε2N−1∑n=0(‖ϕn+1(ϕn+1−ϕn)‖2)=‖σ0u0‖2+λ‖∇ϕ0‖2−λε2‖ϕ0‖2+λ2ε2‖ϕ0‖4L4. |
Proof. Multiplying (2.1b) by 2τλwn+1 and integrating over Ω, we get
2λ(ϕn+1−ϕn,wn+1)+2τλγ‖∇wn+1‖2+2τλ∫Ω˜un+1∇ϕnwn+1dx=0. | (2.5) |
Multiplying (2.1c) by 2λ(ϕn+1−ϕn) yields
2λ(ϕn+1−ϕn,wn+1)=λ2ε2(‖ϕn+1‖4L4−‖ϕn‖4L4)+λ2ε2(‖ϕn+1‖2−‖ϕn‖2)2+λε2‖ϕn+1(ϕn+1−ϕn)‖2+λε2(‖ϕn‖2−‖ϕn+1‖2+‖ϕn+1−ϕn‖2)+λ(‖∇ϕn+1‖2−‖∇ϕn‖2+‖∇(ϕn+1−ϕn)‖2), | (2.6) |
where we use the following identity:
2a(a−b)=a2−b2+(a−b)2,a3(a−b)=14(a4−b4+(a2−b2)2+2a2(a−b)2). | (2.7) |
Testing (2.1a) with τ|˜un+1|2 leads to
‖σn+1˜un+1‖2−‖σn˜un+1‖2+τ∫Ω∇ρn+1⋅un|˜un+1|2dx+τ2∫Ωρn+1(∇⋅un)|˜un+1|2dx=0. | (2.8) |
By taking the inner product of (2.1d) with 2τ˜un+1, we have
‖σn˜un+1‖2−‖σnun‖2+‖σn(˜un+1−un)‖2−2τλ∫Ωwn+1∇ϕn˜un+1dx+2τη‖∇˜un+1‖2+τ∫Ωρn+1(un⋅∇)|˜un+1|2dx+τ2∫Ωρn+1(∇⋅un)|˜un+1|2dx=0. | (2.9) |
Testing (2.1e) with 2τun+1 yields
‖σn+1un+1‖2−‖σn+1˜un+1‖2+‖σn+1(un+1−˜un+1)‖2+ητ‖∇un+1‖2+ητ∣∇˜un+1‖2+ητ‖∇(un+1−˜un+1)‖2=0. | (2.10) |
Taking into account the boundary condition on un and using integration by parts, we have
∫Ω∇ρn+1⋅un|˜un+1|2dx+∫Ωρn+1(∇⋅un)|˜un+1|2dx+∫Ωρn+1un⋅∇|˜un+1|2dx=∫Ω∇⋅(ρn+1un|˜un+1|2)dx=∫Ωρn+1un|˜un+1|2⋅ndx=0. | (2.11) |
Summing the above inequality, we arrive at
2τλγ‖∇wn+1‖2+λ2ε2(‖ϕn+1‖4L4−‖ϕn‖4L4‖)+λ2ε2(‖ϕn+1‖2−‖ϕn‖2)2+λε2‖ϕn+1(ϕn+1−ϕn)‖2+λε2(‖ϕn‖2−‖ϕn+1‖2+‖ϕn+1−ϕn‖2)+λ(‖∇ϕn+1‖2−‖∇ϕn‖2+‖∇(ϕn+1−ϕn)‖2)+‖σn+1un+1‖2−‖σnun‖2+‖σn(˜un+1−un)‖2+‖σn+1(un+1−˜un+1)‖2+ητ‖∇un+1‖2+ητ‖∇˜un+1‖2+ητ‖∇(un+1−˜un+1)‖2=0. | (2.12) |
Adding up the above inequality from n=0 to N−1, we obtain Theorem 2.3.
The bound on the pressure p is proved in the following theorem.
Theorem 3. (Stabilityofp) For any τ>0, the solution pn+1 to (2.1e) satisfies the following inequality:
τ2N−1∑n=0‖pn+1‖2≤C(‖σ0u0‖2+λ‖∇ϕ0‖2−λε2‖ϕ0‖2+λ2ε2‖ϕ0‖4L4)(‖ρ0‖+1). | (2.13) |
Proof. Under the inf-sup condition, there exists a positive constant β such that
β‖pn+1‖≤supv∈V,v≠0(∇⋅v,pn+1)‖∇v‖. | (2.14) |
Testing (2.14) with all v∈V leads to
(∇⋅v,pn+1)=η(∇(un+1−˜un+1),∇v)+τ−1(ρn+1(un+1−˜un+1),v)≤η(‖∇un+1‖+‖∇˜un+1‖)‖∇v‖+τ−1‖σn+1(un+1−˜un+1)‖L2‖σn+1‖L3‖v‖L6. | (2.15) |
Given that ‖ρn‖≤‖ρ0‖ for all 1≤n≤N, and by using Hölder's inequality, we have
‖σn+1‖L3=(∫‖pn+1‖32)13≤C‖ρn+1‖12≤C‖ρ0‖12. | (2.16) |
Then, by the Sobolev embedding inequality ‖v‖L6≤C‖∇v‖L2 for any v∈V, we get
(∇⋅v,pn+1)≤η(‖∇un+1‖+‖∇˜un+1‖)‖∇v‖+Cτ−1‖σn+1(un+1−˜un+1)‖‖ρ0‖12‖∇v‖. | (2.17) |
Substituting the above inequalities into Eq (2.14), we obtain
β‖pn+1‖≤η(‖∇un+1‖+‖∇˜un+1‖)+Cτ−1‖σn+1(un+1−˜un+1)‖‖ρ0‖12. | (2.18) |
And by using Theorem 2.2, we get the desired result.
In this section, we will give the time error estimates and show that the scheme has a first-order convergence rate. Although we verified that the scheme (2.1) is unconditionally stable in the previous chapter, we need to make the following assumptions[34,35] when conducting temporal error analysis:
{ρn}n=0,⋯,N is uniformly bounded in L∞, | (3.1) |
for all n=0,⋯,N, it holds that ρn≥χ a.e. in Ω, | (3.2) |
where χ is a number in (0,ρmin0].
We assume that the exact solution (ρ,u,ϕ,w,p) is sufficiently smooth. To be more precise,
ρ∈H2(0,T;L2(Ω))∩L2(0,T;W1,∞(Ω)),ϕ∈L∞(0,T;H3(Ω))∩W1,∞(0,T;H2(Ω))∩W2,∞(0,T;H1(Ω))∩W3,∞(0,T;L2(Ω)),w∈L∞(0,T;H2(Ω))∩W1,∞(0,T;L2(Ω)),u∈H2(0,T;L2(Ω))∩L∞(0,T;V∩H2(Ω)),p∈W2,∞(0,T;H1(Ω)). | (3.3) |
We denote
enϕ=ϕ(tn)−ϕn,enw=w(tn)−wn,enρ=ρ(tn)−ρn,enu=u(tn)−un,˜enu=u(tn)−˜un,enp=p(tn)−pn. |
In (1.1), taking t=tn+1 and subtracting from (2.1), we get the following error equations:
en+1ρ−enρτ+∇en+1ρ⋅u(tn+1)+∇ρn+1⋅(u(tn+1)−u(tn))+∇ρn+1⋅enu=Rn+1ρ, | (3.4a) |
en+1ϕ−enϕτ−γ△en+1w+u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn=Rn+1ϕ, | (3.4b) |
en+1w+△en+1ϕ=1ε2(ϕ(tn+1)3−(ϕn+1)3−ϕ(tn+1)+ϕn), | (3.4c) |
ρn˜en+1u−enuτ−η△˜en+1u+∇p(tn+1)+ρ(tn+1)(u(tn+1)⋅∇)u(tn+1)−ρn+1(un⋅∇)˜un+1−λw(tn+1)∇ϕ(tn+1)+λwn+1∇ϕn=Rn+1u, | (3.4d) |
ρn+1en+1u−˜en+1uτ−η△(en+1u−˜en+1u)−∇pn+1=0, | (3.4e) |
∇⋅en+1u=0, | (3.4f) |
where
Rn+1ϕ=ϕ(tn+1)−ϕ(tn)τ−ϕt(tn+1),Rn+1ρ=ρ(tn+1)−ρ(tn)τ−ρt(tn+1),Rn+1u=ρnu(tn+1)−u(tn)τ−ρ(tn+1)ut(tn+1). |
If the exact solution is sufficiently smooth, it is easy to establish the following estimate of the truncation error.
Lemma 1. Under the regularity assumptions given by (3.3), the truncation errors satisfy:
‖Rn+1ρ‖2≤Cτ∫tn+1tn‖ρtt(t)‖2dt≤Cτ2,‖Rn+1ϕ‖2≤Cτ∫tn+1tn‖ϕtt(t)‖2dt≤Cτ2,‖Rn+1u‖2≤Cτ∫tn+1tn(‖utt(t)‖2+‖ρt(t)‖2)dt+C‖enρ‖2≤Cτ2+C‖enρ‖2, |
for all 0≤n≤N−1.
Proof. By using the integral residual of the Taylor formula, we have
Rn+1ρ=1τ∫tn+1tn(t−tn)ρtt(t)dt. | (3.5) |
By Hölder's inequality, we can derive
‖Rn+1ρ‖2=∫Ω(1τ∫tn+1tn(t−tn)ρtt(t)dt)2dx≤1τ2∫Ω∫tn+1tn(t−tn)2dt12⋅2∫tn+1tnρtt(t)2dt12⋅2dx≤1τ2(τ33∫tn+1tn∫Ωρtt(t)2dxdt)≤Cτ∫tn+1tn‖ρtt(t)‖2dt≤Cτ2. | (3.6) |
Similarly, we can prove the inequality of Rn+1ϕ. For Rn+1u, we can rewrite
Rn+1u=ρn(u(tn+1)−u(tn)τ−ut(tn+1))−(enρ+∫tn+1tnρt(t)dt)ut(tn+1)=ρnτ∫tn+1tn(t−tn)utt(t)dt−(enρ+∫tn+1tnρt(t)dt)ut(tn+1). | (3.7) |
Using Rn+1ρ estimation and Hölder's inequality can yield the result for Rn+1u.
We introduce the following Gronwall's inequality, which will frequently be used in error estimates.
Lemma 2. Let ak, bk, ck and γk, for integers k≥0, be the nonnegative numbers such that
an+τn∑k=0bk≤τn∑k=0γkak+τn∑k=0ck+B for ≥0. |
Suppose that τγk<1, for all k, and set σk=(1−τγk)−1. Then,
an+τn∑k=0bk≤exp(τn∑k=0γkσk)(τn∑k=0ck+B) for ≥0. |
We verify Lemma 5 by the following lemma.
Lemma 3. Define
Gn+1c=(ϕ(tn+1))3−(ϕn+1)3=3(ϕ(tn+1))2en+1ϕ−3ϕ(tn+1)(en+1ϕ)2+(en+1ϕ)3. | (3.8) |
Then, for n<Tτ−1, we have
‖Gn+1c‖≤C‖en+1ϕ‖H1,‖en+1ϕ‖H2≤C(τ+‖en+1w‖+‖en+1ϕ‖H1+‖enϕ‖). | (3.9) |
Proof. For ‖Gn+1c‖, we use (3.8) to conclude that
‖Gn+1c‖≤C(‖en+1ϕ‖‖ϕ(tn+1)‖2L∞+‖en+1ϕ‖2L4‖ϕ(tn+1)‖L∞+‖en+1ϕ‖3L6)≤C(‖en+1ϕ‖‖ϕ(tn+1)‖2L∞+‖en+1ϕ‖2H1‖ϕ(tn+1)‖L∞+‖en+1ϕ‖3H1)≤C‖en+1ϕ‖H1, | (3.10) |
where we have used the a priori bound ‖ϕn‖H1≤C implied by the stability result given by Theorem 2. Using the H2 regularity results for elliptic equations, we conclude that
‖en+1ϕ‖H2≤c(‖en+1ϕ‖L2+‖△en+1ϕ‖L2); |
from (3.4c), we know that
△en+1ϕ=−en+1w+1ε2Gn+1c−1ε2enϕ−∫tn+1tnϕt(t)dt; |
thus,
‖en+1ϕ‖H2≤C(‖en+1ϕ‖+‖en+1w‖+‖Gn+1c‖+‖enϕ‖+τ)≤C(τ+‖en+1w‖+‖en+1ϕ‖H1+‖enϕ‖). |
The error estimate for the discrete density ρn+1 is derived in the following lemma.
Lemma 4. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) hold. Then, we have
‖en+1ρ‖2+2n∑m=0‖em+1ρ−emρ‖2≤C(τ2+τn∑m=0‖σmemu‖2) | (3.11) |
for all 0≤n≤N−1.
Proof. Multiplying (3.4a) by 2τen+1ρ and integrating over Ω, we have
‖en+1ρ‖2−‖enρ‖2+‖en+1ρ−enρ‖2=−2τ(∇en+1ρ⋅u(tn+1),en+1ρ)+2τ(Rn+1ρ,en+1ρ)−2τ(∇ρn+1⋅(u(tn+1)−u(tn)),en+1ρ)−2τ(∇ρn+1⋅enu,en+1ρ)=4∑i=1Ki. | (3.12) |
Using ∇⋅u(tn+1)=0 in Ω and u(tn+1)=0 on ∂Ω, we have
K1=12∫∂Ω|en+1ρ|2u(tn+1)⋅nds=0. | (3.13) |
By using Young's inequality, the Cauchy-Schwarz inequality and Lemma 1, we have
K2≤2τ‖Rn+1ρ‖‖en+1ρ‖≤Cτ‖Rn+1ρ‖2+ετ‖en+1ρ‖2≤Cτ3+ετ‖en+1ρ‖2, |
where ε>0 is a sufficiently small constant.
Then, by the Sobolev inequality and Young inequality, we get
K3=−2τ(∇ρ(tn+1)⋅(u(tn+1)−u(tn)),en+1ρ)+2τ(∇en+1ρ⋅(u(tn+1)−u(tn)),en+1ρ)=−2τ(∇ρ(tn+1)⋅(u(tn+1)−u(tn)),en+1ρ)≤2τ‖∇ρ(tn+1)‖L∞‖∫tn+1tnut(t)dt‖‖en+1ρ‖≤ετ‖en+1ρ‖2+Cτ2∫tn+1tn‖ut(t)‖2dt≤Cτ3+ετ‖en+1ρ‖2. | (3.14) |
Similarly, the last term can be estimated as follows:
K4≤2τ‖∇ρ(tn+1)‖L∞‖1σn‖L∞‖‖σnenu‖‖en+1ρ‖≤ετ‖en+1ρ‖2+Cτ‖σnenu‖2. | (3.15) |
If ε is sufficiently small such that ετ≤16, substituting the estimates of Ki (1≤i≤4) into (3.12), we have
‖en+1ρ‖2−‖enρ‖2+2‖en+1ρ−enρ‖2≤Cτ3+Cτ‖enρ‖2+Cτ‖σnenu‖2. | (3.16) |
Using the discrete Gronwall inequality, we obtain the desired result.
Lemma 5. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small τ, there are the following error estimates:
τγN−1∑n=0(λ‖∇en+1w‖2+‖en+1w‖2)+‖eNϕ‖2+λ‖∇eNϕ‖2+λ2ε2‖eNϕ‖4L4+‖σNeNu‖2+N−1∑n=0(12‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2)+τηN−1∑n=0(‖∇en+1u‖2+12‖∇˜en+1u‖2+‖∇(en+1u−˜en+1u)‖2)≤Cτ. | (3.17) |
Proof. Let us multiply (3.4a) by τ|˜en+1u|2, (3.4b) by 2τen+1ϕ and 2τλen+1w, (3.4c) by 2τγen+1w and −2λ(en+1ϕ−enϕ), (3.4d) by 2τ˜en+1u and (3.4e) by 2τen+1u. Summing up all of the above equations, we have
2τγλ‖∇en+1w‖2+‖en+1ϕ‖2−‖enϕ‖2+‖en+1ϕ−enϕ‖2+2τγ‖en+1w‖2+λ‖∇en+1ϕ‖2−λ‖∇enϕ‖2+λ‖∇en+1ϕ−enϕ‖2‖σn+1en+1u‖2−‖σnenu‖2+‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2+τη‖∇en+1u‖2+τη‖∇˜en+1u‖2+τη‖∇(en+1u−˜en+1u)‖2=−2τ(ρ(tn+1)(u(tn+1)⋅∇)u(tn+1),˜en+1u)+2τ(ρn+1(un⋅∇)˜un+1,˜en+1u)−τ(∇ρn+1⋅un,|en+1u|2)−2τ(∇p(tn+1),˜en+1u)+2τ(Rn+1u,˜en+1u)+2τλ(w(tn+1)∇ϕ(tn+1)−wn+1∇ϕn,˜en+1u)−2τλ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1w)+2τλ(Rn+1ϕ,en+1w)−2τ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1ϕ)+2τ(Rn+1ϕ,en+1ϕ)+2τγε2(ϕ(tn+1)3−(ϕn+1)3,en+1w)−2τγε2(ϕ(tn+1)−ϕn,en+1w)−2λε2(ϕ(tn+1)3−(ϕn+1)3,en+1ϕ−enϕ)+2λε2(ϕ(tn+1)−ϕn,en+1ϕ−enϕ)=i=14∑i=1Ai. | (3.18) |
Thanks to ∇⋅un=0 in Ω, we have
A2+A3=2τ(ρn+1(un⋅∇)˜un+1,˜en+1u)−τ(∇ρn+1⋅un,|˜en+1u|2)=2τ(ρn+1(un⋅∇)u(tn+1),˜en+1u)−τ(∇⋅(unρn+1|˜en+1u|2),1)=2τ(ρn+1(un⋅∇)u(tn+1),˜en+1u). | (3.19) |
Therefore, we get
A1+A2+A3 | (3.20) |
=2τ(ρn+1(un⋅∇)u(tn+1)−ρ(tn+1)(u(tn+1)⋅∇)u(tn+1),˜en+1u)=−2τ(ρn+1(enu⋅∇)u(tn+1),˜en+1u)−2τ(en+1ρ(u(tn)⋅∇)u(tn+1),˜en+1u)−2τ(ρ(tn+1)((u(tn+1)−u(tn))⋅∇)u(tn+1),˜en+1u)≤Cτ‖ρn+1‖L∞‖1σn‖L∞‖σnenu‖‖∇u(tn+1)‖L3‖˜en+1u‖L6 | (3.21) |
+Cτ‖en+1ρ‖‖u(tn)‖L∞‖∇u(tn+1)‖L3‖˜en+1u‖L6+Cτ‖ρ(tn+1)‖L∞‖∫tn+1tnut(t)dt‖‖∇u(tn+1)‖L3‖˜en+1u‖L6≤Cτ(‖σnenu‖+‖en+1ρ‖+τ∫tn+1tn‖ut(t)‖dt)‖∇˜en+1u‖≤ητ12‖∇˜en+1u‖2+Cτ(‖σnenu‖2+‖en+1ρ‖2+τ2)≤Cτ3+ητ12‖∇˜en+1u‖2+Cτ‖σnenu‖2+Cτ2n∑m=0‖σmemu‖2, | (3.22) |
where the Sobolev embedding inequality ‖v‖L6≤C‖∇v‖L2 for any v∈V is used.
For A4, we have
A4≤2τ‖∇p(tn+1)‖‖1σn‖L∞‖σn(˜en+1u−enu)‖≤Cτ2+12‖σn(˜en+1u−enu)‖2. | (3.23) |
Using the Poincareˊ inequality and Cauchy-Schwarz inequality, we can deal with A_6 , A_7 and A_9 :
\begin{align} A_6 = &2\tau\lambda \left( w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)- w^{n+1} \nabla \phi^{n}, \tilde{e}_{u}^{n+1}\right)\\ = &2\tau\lambda \left( w\left(t_{n+1}\right)\left(\nabla \phi\left(t_{n+1}\right)-\nabla \phi\left(t_{n}\right)\right), \tilde{e}_{u}^{n+1}\right) +2\tau\lambda \left( w\left(t_{n+1}\right)\nabla e_\phi^n, \tilde{e}_{u}^{n+1}\right)\\ &+2\tau\lambda \left( \nabla\phi^n e_w^{n+1}, \tilde{e}_{u}^{n+1}\right)\\ \leq& 2\tau\lambda \left\|w\left(t_{n+1}\right)\right\|_{L^\infty} \left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_tdt\right\|\left\|\tilde{e}_{u}^{n+1}\right\| +2\tau\lambda \left\|w\left(t_{n+1}\right)\right\|_{L^\infty} \left\| \nabla e_\phi^n \right\| \left\|\tilde{e}_{u}^{n+1} \right\|\\ &+2\tau\lambda \left( \nabla\phi^n e_w^{n+1}, \tilde{e}_{u}^{n+1}\right)\\ \leq& C\tau^3+\frac{\eta\tau}{12}\left\|\nabla\tilde{e}_{u}^{n+1}\right\|^2+C\tau\left\| \nabla e_\phi^n \right\|^2+2\tau\lambda \left( \nabla\phi^n e_w^{n+1}, \tilde{e}_{u}^{n+1}\right), \\ A_7 = &-2\tau\lambda \left( \boldsymbol{u}\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)- \tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}, e_{w}^{n+1}\right)\\ = &-2\tau\lambda \left( \boldsymbol{u}\left(t_{n+1}\right)\left(\nabla \phi\left(t_{n+1}\right)-\nabla \phi\left(t_{n}\right)\right) +\boldsymbol{u}\left(t_{n+1}\right)\nabla e_\phi^n+\nabla\phi^n\tilde{e}_{u}^{n+1}, e_{w}^{n+1}\right)\\ \leq& 2\tau\lambda\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^\infty}\left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_tdt\right\| \left\|e_{w}^{n+1}\right\| +2\tau\lambda \left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^\infty} \left\| \nabla e_\phi^n \right\| \left\|e_{w}^{n+1} \right\|\\ & -2\tau\lambda \left(\nabla\phi^n\tilde{e}_{u}^{n+1}, e_{w}^{n+1}\right)\\ \leq& C\tau^3+\frac{\tau\gamma}{3} \left\|e_w^{n+1}\right\|^2+C\tau \left\|\nabla e_\phi^{n}\right\|^2 -2\tau\lambda \left(\nabla\phi^n\tilde{e}_{u}^{n+1}, e_{w}^{n+1}\right), \end{align} | (3.24) |
and
\begin{equation} \begin{split} A_9 = &-2\tau \left( \boldsymbol{u}\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)- \tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}, e_{\phi}^{n+1}\right)\\ \leq& 2\tau \left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^\infty}\left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_tdt\right\| \left\|e_{\phi}^{n+1}\right\| +2\tau \left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^\infty} \left\| \nabla e_\phi^n \right\| \left\|e_{\phi}^{n+1} \right\|\\ & +2\tau\left\|\nabla\phi^n \right\|\left\| \tilde{e}_{u}^{n+1} \right\|_{L_6} \left\| e_{\phi}^{n+1} \right\|_{L_6}\\ \leq& C\tau^3+C\tau \left\|\nabla e_\phi^{n+1}\right\|^2+C\tau \left\|\nabla e_\phi^{n}\right\|^2 +\frac{\eta\tau}{12} \left\|\nabla\tilde{e}_{u}^{n+1}\right\|^2.\\ \end{split} \end{equation} | (3.25) |
From Lemma 1, we have
\begin{equation} \begin{split} A_5+A_8+A_{10} \leq& 2\tau \left\|R_u^{n+1} \right\| \left\| \tilde{e}_{u}^{n+1} \right\| +2\tau\lambda\left\|R_\phi^{n+1} \right\| \left\| e_w^{n+1} \right\| +2\tau \left\| R_\phi^{n+1} \right\| \left\| e_\phi^{n+1} \right\|\\ \leq& C\tau^3+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2} +\frac{\eta\tau}{12} \left\|\nabla \tilde{e}_{u}^{n+1} \right\|^2+\frac{\tau\gamma}{3} \left\| e_w^{n+1} \right\|^2 +\frac{1}{2} \left\| e_\phi^{n+1} \right\|^2.\\ \end{split} \end{equation} | (3.26) |
From Lemma 3, we obtain
\begin{equation} \begin{aligned} A_{11}+A_{12} & = \frac{2\tau\gamma}{\varepsilon^2}\left(G_{c}^{n+1}-e_{\phi}^n-\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, e_{w}^{n+1}\right) \\ & \leq \frac{2\tau\gamma}{\varepsilon^2}\left(\left\|e_{\phi}^n\right\|+\left\|G_{c}^{n+1}\right\|+ \left\|\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt\right\|\right)\left\|e_{w}^{n+1}\right\| \\ & \leq C\tau^3+C \tau\left(\left\|e_{\phi}^{n}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}\right) +\frac{\tau\gamma}{3}\left\|e_{w}^{n+1}\right\|^{2}, \\ \end{aligned} \end{equation} | (3.27) |
and
\begin{align} A_{13}+A_{14} = &- \frac{2\lambda}{\varepsilon^2}\left(G_{c}^{n+1}-e_{\phi}^n-\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, e_{\phi}^{n+1}-e_{\phi}^{n}\right) \\ = &-\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|\left(e_{\phi}^{n+1}\right)^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e_{\phi}^{n}\right)\right\|^{2}\right)-\frac{2\lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^{2} \\ &-\frac{2\lambda}{\varepsilon^{2}}\left(3\left(\phi\left(t_{n+1}\right)\right)^{2} e_{\phi}^{n+1}-3 \phi\left(t_{n+1}\right)\left(e_{\phi}^{n+1}\right)^{2}-e_{\phi}^{n+1}, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\\ &+\frac{2\lambda}{\varepsilon^{2}} \left(\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, e_{\phi}^{n+1}-e_{\phi}^{n}\right), \end{align} | (3.28) |
where we use this identity
\begin{equation} \begin{split} a^{3}(a-b) = \frac{1}{4}\left(a^{4}-b^{4}+\left(a^{2}-b^{2}\right)^{2}+2 a^{2}(a-b)^{2}\right). \end{split} \end{equation} | (3.29) |
We denote
\begin{equation} \tilde{G}^{n+1} = -\frac{2\lambda}{\varepsilon^{2}}\left(3\left(\phi\left(t_{n+1}\right)\right)^{2} e_{\phi}^{n+1}-3 \phi\left(t_{n+1}\right)\left(e_{\phi}^{n+1}\right)^{2}-e_{\phi}^{n+1}\right) \text {. } \end{equation} | (3.30) |
Similar to the method estimated from Lemma 3, we can obtain
\begin{equation} \left\|\tilde{G}^{n+1}\right\| \leq C\left\|e_{\phi}^{n+1}\right\|_{H^{1}} . \end{equation} | (3.31) |
Taking the gradient of \tilde{G}^{n+1} , we get
\begin{equation} \begin{aligned} \nabla \tilde{G}^{n+1} = &-\frac{2\lambda}{\varepsilon^{2}}\left[\left(3\left(\phi\left(t_{n+1}\right)\right)^{2}-1\right) \nabla e_{\phi}^{n+1}+6 \phi\left(t_{n+1}\right) e_{\phi}^{n+1} \nabla \phi\left(t_{n+1}\right)\right.\\ &\left.-3\left(e_{\phi}^{n+1}\right)^{2} \nabla \phi\left(t_{n+1}\right)-6 \phi\left(t_{n+1}\right) e_{\phi}^{n+1} \nabla e_{\phi}^{n+1}\right]. \end{aligned} \end{equation} | (3.32) |
Since H^{2}(\Omega) \subset L^{\infty}(\Omega) and by utilizing the bound of \left\|\nabla e_{\phi}^{n+1}\right\|_{L^{2}} implied by Theorem 2, we conclude that
\begin{equation} \left\|e_{\phi}^{n+1} \nabla e_{\phi}^{n+1}\right\|_{L^{2}} \leq\left\|e_{\phi}^{n+1}\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n+1}\right\|_{L^{2}} \leq C\left\|e_{\phi}^{n+1}\right\|_{H^{2}} . \end{equation} | (3.33) |
In view of (3.10) and the bound \left\|\phi^{n}\right\|_{H^{1}} < C , we have
\begin{equation} \begin{array}{rl} \left\|\nabla \tilde{G}^{n+1}\right\| \leq & C\left[\left(\left\|\phi\left(t_{n+1}\right)\right\|_{L^{\infty}}^{2}+1\right)\left\|\nabla e_{\phi}^{n+1}\right\|_{L^{2}}\right.\\ &+\left\|\phi\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla \phi\left(t_{n+1}\right)\right\|_{L^{3}}\left\|e_{\phi}^{n+1}\right\|_{L^{6}} \\ & \left.+\left\|\nabla \phi\left(t_{n+1}\right)\right\|_{L^{6}}\left\|e_{\phi}^{n+1}\right\|_{L^{6}}^{2}+\left\|\phi\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|e_{\phi}^{n+1} \nabla e_{\phi}^{n+1}\right\|_{L^{2}}\right] \\ \leq &C\left(\left\|\nabla e_{\phi}^{n+1}\right\|_{L^{2}}+\left\|e_{\phi}^{n+1}\right\|_{H^{1}}\right. \left.+\left\|e_{\phi}^{n+1}\right\|_{H^{1}}^{2}+\left\| e_{\phi}^{n+1} \nabla e_{\phi}^{n+1} \right\|_{L^{2}}\right) \\ \leq &C\left(\tau+\left\|e_{w}^{n+1}\right\|+\left\|e_{\phi}^{n+1}\right\|+\left\|\nabla e_{\phi}^{n+1}\right\| +\left\|e_{\phi}^{n}\right\| \right) . \end{array} \end{equation} | (3.34) |
Dealing with the penultimate term of (3.28) and in view of (3.31) and (3.33), we get
\begin{equation} \begin{split} &\left(\tilde{G}^{n+1}, e_{\phi}^{n+1}-e_{\phi}^{n}\right) \\ = &\tau\left(\tilde{G}^{n+1}, \gamma \Delta e_{w}^{n+1}-\boldsymbol{u}\left(t_{n+1}\right) \nabla e_{\phi}^{n}-\tilde{e}_{u}^{n+1} \nabla \phi^{n}\right)\\ &+\tau\left(\tilde{G}^{n+1}, -\boldsymbol{u}\left(t_{n+1}\right)\int_{t_{n}}^{t_{n+1}}\nabla\phi_t\left(t\right)dt+R_{\phi}^{n+1}\right)\\ \leq& \gamma\tau\left\|\nabla e_{w}^{n+1}\right\|\left\|\nabla \tilde{G}^{n+1}\right\| +\tau\left\| \nabla\phi^n\right\|\left\| \tilde{e}_u^{n+1}\right\|_{H_1}\left\| \tilde{G}^{n+1} \right\|_{H_1}\\ &+\tau\left\|\tilde{G}^{n+1}\right\|\left(\left\|R_{\phi}^{n+1}\right\|+\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n}\right\|+\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_t\left(t\right)dt\right\| \right) \\ \leq& C\tau^3+\frac{\gamma\lambda\tau}{2}\left\|\nabla e_{w}^{n+1}\right\|^2+\frac{\eta\tau}{12}\left\| \nabla\tilde{e}_u^{n+1}\right\|^2\\ &+C\tau\left(\left\|e_{w}^{n+1}\right\|^2+\left\|e_{\phi}^{n+1}\right\|^2+\left\|\nabla e_{\phi}^{n+1}\right\|^2 +\left\|e_{\phi}^{n}\right\| ^2+\left\|\nabla e_{\phi}^{n}\right\| ^2 \right).\\ \end{split} \end{equation} | (3.35) |
Then, we estimate the last term of (3.28),
\begin{align} & \frac{2\lambda}{\varepsilon^2}\left(\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, e_{\phi}^{n+1}-e_{\phi}^{n}\right) \\ = &\frac{2\lambda\tau}{\varepsilon^2}\left(\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, \gamma \Delta e_{w}^{n+1}-\boldsymbol{u}\left(t_{n+1}\right) \nabla e_{\phi}^{n}-\tilde{e}_{u}^{n+1} \nabla \phi^{n}\right)\\ &+\frac{2\lambda\tau}{\varepsilon^2}\left(\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt, -\boldsymbol{u}\left(t_{n+1}\right)\int_{t_{n}}^{t_{n+1}}\nabla\phi_t\left(t\right)dt+R_{\phi}^{n+1}\right)\\ \leq& \frac{2\lambda\gamma\tau}{\varepsilon^2}\left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_t\left(t\right)dt\right\|\left\|\nabla e_w^{n+1}\right\| +\frac{2\lambda\tau}{\varepsilon^2}\left\| \nabla\phi^n\right\|\left\| \tilde{e}_u^{n+1}\right\|_{H_1}\left\| \int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt \right\|_{H_1}\\ &+\frac{2\lambda\tau}{\varepsilon^2}\left\|\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt\right\|\left(\left\|R_{\phi}^{n+1}\right\|+\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n}\right\| \right) \\ &+\frac{2\lambda\tau}{\varepsilon^2}\left\|\int_{t_{n}}^{t_{n+1}}\phi_t\left(t\right)dt\right\|\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\int_{t_{n}}^{t_{n+1}}\nabla\phi_t\left(t\right)dt\right\| \\ \leq& C\tau^3+\frac{\gamma\lambda\tau}{2}\left\|\nabla e_{w}^{n+1}\right\|^2+\frac{\eta\tau}{12}\left\|\nabla \tilde{e}_{u}^{n+1}\right\| ^2 +C\tau\left\|\nabla e_{\phi}^{n}\right\| ^2 .\\ \end{align} | (3.36) |
Combining (3.35) and (3.36), we can obtain
\begin{equation} \begin{split} A_{13}+A_{14} \leq&-\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|\left(e_{\phi}^{n+1}\right)^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e_{\phi}^{n}\right)\right\|^{2}\right)-\frac{2\lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^{2} \\ &+ C\tau^3+\gamma\lambda\tau\left\|\nabla e_{w}^{n+1}\right\|^2+\frac{\eta\tau}{6}\left\|\nabla \tilde{e}_{u}^{n+1}\right\| ^2\\ &+C\tau \left( \left\|\nabla e_{\phi}^{n}\right\| ^2+\left\|e_{w}^{n+1}\right\|^2+\left\|e_{\phi}^{n+1}\right\|^2+\left\|\nabla e_{\phi}^{n+1}\right\|^2 +\left\|e_{\phi}^{n}\right\| ^2+\left\|\nabla e_{\phi}^{n}\right\| ^2 \right) .\\ \end{split} \end{equation} | (3.37) |
Substituting the above estimates into (3.18), we have
\begin{equation} \begin{split} &\tau\gamma\lambda\left\| \nabla e_{w}^{n+1} \right\|^2+ \frac{1}{2}\left\| e_{\phi}^{n+1}\right\|^2- \left\| e_{\phi}^{n}\right\|^2+ (1+\frac{2\lambda}{\varepsilon^{2}})\left\| e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^2\\ &+\tau\gamma\left\| e_{w}^{n+1} \right\|^2+\lambda\left\| \nabla e_{\phi}^{n+1}\right\|^2- \lambda\left\| \nabla e_{\phi}^{n}\right\|^2+ \lambda\left\| \nabla e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^2\\ &+\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}-\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\frac{1}{2}\left\|\sigma^{n}\left(\tilde{e}_{u}^{n+1}-e_{u}^{n}\right)\right\|^{2}+\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\|^{2}\\ &+ \tau\eta\left\|\nabla e_{u}^{n+1}\right\|^{2}+ \frac{\tau\eta}{2}\left\|\nabla \tilde{e}_{u}^{n+1}\right\|^{2}+ \tau\eta\left\|\nabla\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\|^{2} \\ &+\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|\left(e_{\phi}^{n+1}\right)^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e_{\phi}^{n}\right)\right\|^{2}\right)\\ \leq& C \tau^{2}+C \tau\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}\\ &+C\tau \left( \left\|\nabla e_{\phi}^{n}\right\| ^2+\left\|e_{w}^{n+1}\right\|^2+\left\|e_{\phi}^{n+1}\right\|^2+\left\|\nabla e_{\phi}^{n+1}\right\|^2 +\left\|e_{\phi}^{n}\right\| ^2+\left\|\nabla e_{\phi}^{n}\right\| ^2 \right).\\ \end{split} \end{equation} | (3.38) |
Adding up from 0 to N-1 , and applying Gronwall's inequality, we infer that
\begin{align*} \label{eq12} &\tau\gamma\sum\limits_{n = 0}^{N-1}\left(\lambda\left\| \nabla e_{w}^{n+1} \right\|^2+\left\| e_{w}^{n+1} \right\|^2 \right)+\left\| e_{\phi}^{N}\right\|^2+\lambda\left\| \nabla e_{\phi}^{N}\right\|^2+\frac{\lambda}{2 \varepsilon^{2}}\left\|e_{\phi}^{N}\right\|_{L^{4}}^{4}\\ &+\left\|\sigma^{N} e_{u}^{N}\right\|^{2}+ \sum\limits_{n = 0}^{N-1}\left(\frac{1}{2}\left\|\sigma^{n}\left(\tilde{e}_{u}^{n+1}-e_{u}^{n}\right)\right\|^{2}+\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\|^{2}\right)\\ &+\tau\eta\sum\limits_{n = 0}^{N-1}\left(\left\|\nabla e_{u}^{n+1}\right\|^{2}+ \frac{1}{2}\left\|\nabla \tilde{e}_{u}^{n+1}\right\|^{2}+ \left\|\nabla\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\|^{2} \right)\\ \leq& C\tau. \end{align*} |
Lemma 5 shows that the fractional scheme converges at a rate of O(\tau^{\frac{1}{2}}) . However, since we use the first-order backward Euler method of time discretization, it is not optimal from the perspective of theoretical analysis. Next, we will improve the convergence speed to the first order.
Lemma 6. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small \tau , there are the following error estimates:
\begin{equation} \begin{split} &\tau\gamma\sum\limits_{n = 0}^{N-1}\left(\lambda\left\| \nabla e_{w}^{n+1} \right\|^2+\left\| e_{w}^{n+1} \right\|^2 \right)+\left\| e_{\phi}^{N}\right\|^2+\lambda\left\| \nabla e_{\phi}^{N}\right\|^2\\ &+\frac{\lambda}{2 \varepsilon^{2}}\left\|e_{\phi}^{N}\right\|_{L^{4}}^{4} +\left\|\sigma^{N} e_{u}^{N}\right\|^{2}+\tau\eta\sum\limits_{n = 0}^{N-1}\left\|\nabla e_{u}^{n+1}\right\|^{2}\\ \leq& C\tau^2.\\ \end{split} \end{equation} | (3.39) |
Proof. Taking the sum of (3.4d) and (3.4e), we get
\begin{equation} \begin{split} &\rho^{n} \frac{e_u^{n+1}-e_u^ n}{\tau}+\frac{\rho^{n+1}-\rho^{n}}{\tau}\left(e_{u}^{n+1}-\tilde{e}_{n}^{n+1}\right)-\eta \triangle e_{u}^{n+1}+\rho\left(t_{n+1}\right)\left(u\left(t_{n+1}\right) \cdot \nabla\right) u(t_{n+1})\\ &-\rho^{n+1}\left(u^{n} \cdot \nabla\right) \tilde{u}^{n+1}+\Delta e_{p}^{n+1}-\lambda w\left(t _{n+1}\right) \nabla \phi(t_{n+1})+\lambda w^{n+1}\nabla\phi^{n} = R_{u}^{n+1}.\\ \end{split} \end{equation} | (3.40) |
Let us multiply (3.4a) by \tau\tilde{e}_{u}^{n+1} , (3.4b) by 2 \tau e_{\phi}^{n+1} and 2 \tau \lambda e_{w}^{n+1} , (3.4c) by 2\tau\gamma e_{w}^{n+1} and -2 \lambda(e_{\phi}^{n+1}-e_{\phi}^{n}) and (3.40) by 2 \tau e_{u}^{n+1} . Summing up all of the above equations, we obtain
\begin{align} &\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}-\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\left\|\sigma^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right)\right\|^{2}+ 2 \eta\tau\left\|\nabla e_{u}^{n+1}\right\|^{2}+2 \tau \gamma \lambda\left\|\nabla e_{w}^{n+1}\right\|^{2}\\ &+\left\|e_{\phi}^{n+1}\right\|^{2}-\left\|e_{\phi}^{n}\right\|^{2}+\left\|e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^{2}+2 \tau \gamma\left\|e_{w}^{n+1}\right\|^{2}+\lambda\left\|\nabla e_{\phi}^{n+1}\right\|^{2}-\lambda\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+\lambda\left\|\nabla e _{\phi}^{n+1}-\nabla e_{\phi}^{n} \right\|^{2}\\ = &-2\tau\left(\nabla\rho^{n+1}\cdot u^n, \tilde{e}_{u}^{n+1}\cdot e_{u}^{n+1} \right)+\tau\left( \nabla\rho^{n+1}\cdot u^n, \vert e_{u}^{n+1}\vert^2 \right)\\ &-2 \tau\left(\rho\left(t_{n+1}\right)\left(u\left(t_{n+1}\right)\cdot\nabla\right) u\left(t_{n+1}\right)-\rho^{n+1}\left(u^{n}\cdot\nabla\right) \tilde{u}^{n+1}, e_u^{n+1}\right)\\ &+2 \tau \lambda\left(w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-w^{n+1} \nabla \phi^{n}, e_{u}^{n+1}\right)+2 \tau\left(R_{u}^{n+1}, e_u^{n+1}\right)\\ &-2 \tau \lambda\left(u\left(t_{n+1}\right) \nabla \phi\left(t _{n+1}\right)-\tilde{u}^{n+1} \nabla \phi^{n}, e_{w}^{n+1}\right)+2 \tau \lambda \left(R_{\phi}^{n+1}, e_w^{n+1}\right)\\ &-2 \tau\left(u\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-\tilde{u}^{n+1} \nabla \phi^{n}, e_{\phi}^{n+1}\right)+2 \tau\left(R_{\phi}^{n+1}, e_{\phi}^{n+1}\right)\\ &+\frac{2 \tau r}{\varepsilon^{2}}\left(\phi\left(t_{n+1}\right)^{3}-\left(\phi^{n+1}\right)^{3}, e_w^{n+1}\right)-\frac{2 \tau r}{\varepsilon^{2}}\left(\phi\left(t_{n+1}\right)-\phi^{n}, e_{w}^{n+1}\right)\\ &-\frac{2 \lambda}{\varepsilon^{2}}\left(\phi\left(t_{n+1}\right)^{3}-\left(\phi^{n+1}\right)^{3}, e_{\phi}^{n+1}-e_{\phi}^{n}\right)+\frac{2 \lambda}{\varepsilon^{2}}\left(\phi\left(t_{n+1}\right)-\phi^{n}, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\\ = &\sum\limits_{i = 1}^{13} L_{i}. \end{align} | (3.41) |
Using integration by parts, we have
\begin{equation} \begin{split} L_{1}+L_{2} = & 2 \tau\left(\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde{e}_{u}^{n+1}, e_{ u}^{n+1}\right)+2 \tau\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) e_{u}^{n+1}, e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right) \\ &-2 \tau\left(\rho^{n+1}\left(\boldsymbol{u}\left(t_{n}\right) \cdot \nabla\right) e_{u}^{n+1}, e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right){;} \end{split} \end{equation} | (3.42) |
we rewrite the L_{3} as
\begin{equation} \begin{split} L_{3} = &-2 \tau\left(\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde e_{u}^{n+1}, e_{u}^{n+1}\right)-2 \tau\left(\rho^{n+1}\left(\left(\boldsymbol{u}\left(t_{n+1}\right)-\boldsymbol{u}\left(t_{n}\right)\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right) \\ &-2 \tau\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right)-2 \tau\left(e_{\rho}^{n+1}\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right){;} \end{split} \end{equation} | (3.43) |
then,
\begin{equation} \begin{split} L_{1}+L_{2}+L_{3} = & 2 \tau\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) e_{u}^{n+1}, e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right) \\ &-2 \tau\left(e_{\rho}^{n+1}\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right) \\ &-2 \tau\left(\rho^{n+1}\left(\boldsymbol{u}\left(t_{n}\right) \cdot \nabla\right) e_{u}^{n+1}, e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right) \\ &-2 \tau\left(\rho^{n+1}\left(\left(\boldsymbol{u}\left(t_{n+1}\right)-\boldsymbol{u}\left(t_{n}\right)\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right) \\ &-2 \tau\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), e_{u}^{n+1}\right) \\ = & \sum\limits_{i = 1}^{5} J_{i}. \\ \end{split} \end{equation} | (3.44) |
Lemma 5 shows that
\begin{equation} \begin{split} \left\|\nabla\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\| \leq C.\quad\quad \end{split} \end{equation} | (3.45) |
According to Young's inequality and the Cauchy-Schwarz inequality, we can deduce that
\begin{align} J_{1} & \leq \frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{u}^{n}\right\|^{2}\left\|\nabla\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\|\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\| \\ & \leq \frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau^{\frac{3}{2}}\left\|\nabla e_{u}^{n}\right\|^{2}, \\ J_{2} & \leq 2 \tau\left\|e_{\rho}^{n+1}\right\|\left\|\boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla \boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{3}}\left\|e_{u}^{n+1}\right\|_{L^{6}} \\ & \leq \frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau\left\|e_{\rho}^{n+1}\right\|^{2} \\ & \leq C \tau^{3}+\frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}, \\ J_{3} & \leq C \tau\left\|\nabla e_{u}^{n+1}\right\|\left\|\boldsymbol{u}\left(t_{n}\right)\right\|_{{L}^{\infty}}\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\tilde{e}_{u}^{n+1}\right)\right\| \\ & \leq C \tau^{3}+\frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2, \\ J_{4} & \leq 2 \tau\left\|\rho^{n+1}\right\|_{{L}^{\infty}}\left\|\nabla \boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\int_{t_{n}}^{t_{n+1}} \boldsymbol{u}_{t} d t\right\|\left\|e_{u}^{n+1}\right\| \\ & \leq C \tau^{3}+\frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}, \\ J_{5} & \leq 2 \tau\left\|\rho^{n+1}\right\|_{L^{\infty}}\left\|\nabla \boldsymbol{u}\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\frac{1}{\sigma^{n}}\right\|_{L^{\infty}}\left\|\sigma^{n} e_{u}^{n}\right\|\left\|e_{u}^{n+1}\right\| \\ & \leq C \tau\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\frac{\eta \tau}{10}\left\|\nabla e_{u}^{n+1}\right\|^{2}. \end{align} | (3.46) |
Therefore, we have
\begin{equation} \begin{split} M_{1}+M_{2}+M_{3} \leq& C \tau^{3}+C \tau^{\frac{3}{2}}\left\|\nabla e_{u}^{n}\right\|^{2}+\frac{\eta \tau}{2}\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau\left\|\sigma^{n} e_{u}^{n}\right\|^{2}\\ &+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2. \end{split} \end{equation} | (3.47) |
Using the embedding inequality and Cauchy-Schwarz inequality, we can deal with M_4 – M_9
\begin{equation} \begin{split} M_{4} = &2 \tau \lambda\left(w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-w^{n+1} \nabla \phi^{n}, e_{u}^{n+1}\right)\\ = &2 \tau \lambda\left(w\left(t_{n+1}\right)\left(\nabla \phi\left(t_{n+1}\right)-\nabla \phi\left(t_{n}\right)\right)+w\left(t_{n+1}\right) \nabla e_{\phi}^{n}+\nabla \phi^{n} e_{w}^{n+1}, e_{u}^{n+1}\right)\\ \leq& 2 \tau \lambda \left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\| \int_{t_{n}}^{t_{n+1}} \nabla \phi_{t}(t) d t\right\|\| e_{u}^{n+1}\|+2 \tau \lambda\left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n}\right\|\left\|e_{u}^{n+1}\right\|\\ &+2 \tau \lambda\left\|\nabla \phi^{n}\right\|_{L^2}\left\|e_w^{n+1}\right\|_{L^{3}}\left\|e_u^{n+1}\right\|_{L^{6}}\\ \leq& C \tau^{3}+\tau \eta\left\|\nabla e_{u}^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{\phi}^{n}\right\|^{2}+C \tau\left\|\nabla e_{w}^{n+1}\right\|^{2}, \\ \end{split} \end{equation} | (3.48) |
\begin{equation} \begin{split} M_{5}+M_{7}+M_{9}& = 2 \tau\left(R_{u}^{n+1}, e_{u}^{n+1}\right)+2\tau \lambda\left(R_{\phi}^{n+1}, e_{w}^{n+1}\right)+2\tau\left(R_{\phi}^{n+1}, e_{\phi}^{n+1}\right)\\ &\leq2\tau\left\|R_{u}^{n+1}\right\|\left\|e_{u}^{n+1}\right\|+2 \tau \lambda\left\|R_{\phi}^{n+1}\right\|\left\|e_{w}^{n+1}\right\|+2\tau\left\|R_{\phi}^{n+1}\right\|\left\|e_{\phi}^{n+1}\right\|\\ &\leq C \tau^{3}+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}+\frac{\tau\lambda\gamma}{4}\left\|\nabla e_w^{n+1}\right\|^{2}+\frac{1}{4}\| e_\phi^{n+1} \|^{2}, \end{split} \end{equation} | (3.49) |
\begin{equation} \begin{split} M_{6} = &-2 \tau \lambda\left( \boldsymbol{u}\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-\tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}, e_{w}^{n+1}\right)\\ = &-2 \tau \lambda\left(\left(\tilde{e}_{u}^{n+1}-e_{u}^{n+1}\right) \nabla \phi\left(t_{n+1}\right), e_w^{n+1}\right)-2 \tau\lambda\left(e_{u}^{n+1} \nabla \phi\left(t_{n+1}\right), e_{w}^{n+1}\right)\\ &-2 \tau \lambda\left(\tilde{\boldsymbol{u}}^{n+1} \int_{t_n}^{t_{n+1}} \nabla \phi_t(t) d t, e_{w}^{n+1}\right)-2 \tau \lambda \left(\tilde{\boldsymbol{u}}^{n+1} \nabla e_{\phi}^{n}, e_{w}^{n+1}\right)\\ \leq& 2 \tau \lambda\left\|\frac{1}{\sigma^{n+1}}\right\|_{L^{\infty}}\left\|\sigma^{n+1}\left(\tilde{e}_{u}^{{n+1}}-e_{u}^{n+1}\right)\right\|\left\|\nabla \phi\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|e_{w}^{n+1}\right\|\\ &+2\tau \lambda\left\|\frac{1}{\sigma^{n+1}}\right\|_{L^{\infty}}\left\|\sigma^{n+1} e_{u}^{n+1}\right\|\left\|\nabla \phi\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|e_w^{n+1}\right\|\\ &+2 \tau \lambda\left\|\int_{t_{n}}^{t_{n+1}} \nabla \phi_{t}(t) d t\right\|_{L^{2}}\left\|\tilde{\boldsymbol{u}}^{n+1}\right\|_{L^{3}}\left\|e_{w}^{n+1}\right\|_{L^{6}}\\ &+2 \tau \lambda\left\|\nabla e_{\phi}^{n}\right\|\left\|\tilde{\boldsymbol{u}}^{n+1}\right\| _{L^{3}}\left\|e_{w}^{n+1}\right\|_{L^{6}}\\ \leq& C \tau^{3}+\frac{\tau\lambda}{2}\left\|e_w^{n+1}\right\|^{2}+C \tau\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}+\frac{\tau\lambda\gamma}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2, \\ \end{split} \end{equation} | (3.50) |
and
\begin{equation} \begin{split} M_8 & = -2 \tau\left(u\left(t_{n+1}\right) \nabla \phi(t_{n+1} )-\tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}, e_{\phi}^{n+1}\right) \\ & \leq C\tau^{3}+\frac{1}{4}\left\|e_{\phi}^{n+1}\right\|^{2}+C\tau\left\|\sigma^{n+1} e_u^{n+1}\right\|^{2}+\frac{\lambda}{2}\left\|\nabla e_{\phi}^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{\phi}^{n}\right\|^{2}.\\ \end{split} \end{equation} | (3.51) |
Next, we estimate M_{10} + M_{11} and M_{12} + M_{13} as follows:
\begin{equation} \begin{split} M_{10}+M_{11} \leq C\tau^{3}+C \tau\left(\left\|e_{\phi}^{n}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}\right)+\frac{\tau\gamma}{2} \left\|e_{w}^{n+1}\right\|^{2}. \end{split} \end{equation} | (3.52) |
\begin{align} &\left(\tilde{G}^{n+1}, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\\ = &\tau\left(\tilde{G}^{n+1}, \gamma \triangle e_w^{n+1}-u(t_{n+1} ) \nabla \phi(t _{n+1})+\tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}+R_{\phi}^{n+1}\right)\\ = &\tau\left(\tilde{G}^{n+1}, \gamma \triangle e_w^{n+1}-\left(\tilde{e}_{u}^{n+1}-e_{u}^{n+1}\right) \nabla \phi\left(t_{n+1}\right)-e_{u}^{n+1} \nabla \phi(t _{n+1})-\tilde{\boldsymbol{u}}^{n+1} \int_{t _n}^{t _{n+1}} \nabla \phi_t(t) d t\right)\\ &-\tau\left(\tilde{G}^{n+1}, \tilde{\boldsymbol{u}} \nabla e_{\phi}^{n}-R_{\phi}^{n+1}\right) \\ \leq &C \tau^{3}+C\tau\left\|\tilde{G}^{n+1}\right\|^{2}+\frac{\tau\gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+C\tau\left\|\nabla \tilde{G}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e_u^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ \leq& C\tau^{3}+\frac{\tau \gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ &+C\tau\left(\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}\right), \end{align} | (3.53) |
and
\begin{align} \left(\int_{t_{n}}^{t_{n+1}} \phi_{t}(t) d t, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\leq C \tau^{3}+\frac{\tau \gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e^{n+1}\right\|^{2}+C \tau\|\nabla e _{\phi}^{n}\|^{2}. \end{align} |
From (3.29), combining the two inequalities above, we deduce that
\begin{equation} \begin{split} M_{12}+M_{13} \leq&-\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|(e_{\phi}^{n+1})^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e _{\phi}^{n}\right)\right\|^{2}\right)-\frac{2 \lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e _{\phi}^{n}\right\|^{2}\\ &+C \tau^{3}+\frac{\tau\gamma \lambda}{2}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{2}\left\|\sigma^{n+1} e_u^{n+1}\right\|^{2} \\ &+C\tau\left(\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}\right).\\ \end{split} \end{equation} | (3.54) |
Plugging the above inequality into (3.41) gives
\begin{align} &\frac{1}{2}\left(\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}-\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\left\|\sigma^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right)\right\|^{2}\right)+ \eta\tau\left\|\nabla e_{u}^{n+1}\right\|^{2}+ \tau \gamma \lambda\left\|\nabla e_{w}^{n+1}\right\|^{2} \\ &+\frac{1}{2}\left(\left\|e_{\phi}^{n+1}\right\|^{2}-\left\|e_{\phi}^{n}\right\|^{2}+\left\|e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^{2}\right)+ \tau \gamma\left\|e_{w}^{n+1}\right\|^{2}+\frac{\lambda}{2}\left\|\nabla e_{\phi}^{n+1}\right\|^{2}-\lambda\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+\lambda\left\|\nabla e _{\phi}^{n+1}-\nabla e_{\phi}^{n} \right\|^{2}+\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|(e_{\phi}^{n+1})^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e _{\phi}^{n}\right)\right\|^{2}\right)+\frac{2 \lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e _{\phi}^{n}\right\|^{2}\\ \leq& C \tau^{3}+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ &+C\tau\left(\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{w}^{n+1}\right\|^{2} +\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}\right){;} \end{align} | (3.55) |
for sufficiently small \tau , taking the sum of (3.55) from 0 to N -1 and using the discrete Gronwall inequality, we can obtain Lemma 6.
Theorem 4. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small \tau , there are the following error estimates:
\begin{equation} \begin{aligned} &\left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2}+\left\|\rho\left(t_{n}\right)-\rho^{n}\right\|^{2}+\left\|\nabla\left(w\left(t_{n}\right)-w^{n}\right)\right\|^{2} \\ &+\left\|\nabla\left(\phi\left(t_{n}\right)-\phi^{n}\right)\right\|^{2}+\eta \tau \sum\limits_{m = 1}^{n}\left\|\nabla\left(\boldsymbol{u}\left(t_{m}\right)-\boldsymbol{u}^{m}\right)\right\|^{2} \leq C \tau^{2} \end{aligned} \end{equation} | (3.56) |
for all 1 \leq n \leq N .
Proof. From Lemmas 5 and 6, we obtain
\begin{equation} \left\|\rho\left(t_{n}\right)-\rho^{n}\right\|^{2}+\left\|\nabla\left(w\left(t_{n}\right)-w^{n}\right)\right\|^{2}+\left\|\nabla\left(\phi\left(t_{n}\right)-\phi^{n}\right)\right\|^{2} +\eta \tau \sum\limits_{m = 1}^{n}\left\|\nabla\left(\boldsymbol{u}\left(t_{m}\right)-\boldsymbol{u}^{m}\right)\right\|^{2} \leq C \tau^{2}. \end{equation} | (3.57) |
Thus, we only prove that
\begin{equation} \left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2} \leq C \tau^{2} . \end{equation} | (3.58) |
In fact, we have
\begin{equation} \begin{aligned} \left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2} & \leq C\left\|\left(\sigma\left(t_{n}\right)-\sigma^{n}\right)\boldsymbol{u}\left(t_{n}\right)\right\|^{2}+\left\|\sigma^{n} e_{u}^{n}\right\|^{2} \\ & \leq C\left(\left\|e_{\rho}^{n}\right\|^{2}+\left\|\sigma^{n} e_{u}^{n}\right\|^{2}\right) \leq C \tau^{2}. \end{aligned} \end{equation} | (3.59) |
Theorem 3.6 states that both \sigma^{n}\boldsymbol{u}^{n}, \; \rho^{n}, \; \boldsymbol{u}^{n} , \phi^{n} and w^{n} are order 1 approximations to \sigma\boldsymbol{u}, \; \rho, \; \boldsymbol{u} , \phi and w in l^{\infty}\left(L^{2}(\Omega)\right) , l^{\infty}\left(L^{2}(\Omega)\right) , l^{2}\left(H_{0}^{1}(\Omega)\right) , l^{\infty}\left(H_0^{1}(\Omega)\right) and l^{\infty}\left(H_0^{1}(\Omega)\right) , respectively. Finally, we can obtain order \frac{1}{2} error estimates for p approximation in l^{\infty}\left(L^{2}(\Omega)\right) .
Theorem 5. Under the assumptions in Theorem 4, the following holds true:
\begin{equation} \tau\sum\limits_{m = 1}^{N}\left\|p\left(t_{m}\right)-p^{m}\right\| \leq C \tau. \end{equation} | (3.60) |
Proof. Let us rewrite (3.40) as
\begin{equation} \begin{split} -\nabla e_{p}^{n+1} = &\rho^{n} \frac{e_{u}^{n+1}-e_{u}^{n}}{\tau}+\frac{\rho^{n+1}-\rho^{n}}{\tau}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)-\eta \triangle e_{u}^{n+1}-R_{u}^{n+1}\\ &+\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde{\boldsymbol{u}}^{n+1}\\ &- \lambda w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)+ \lambda w^{n+1} \nabla \phi^{n}.\\ \end{split} \end{equation} | (3.61) |
To prove the theorem, we introduce the discrete inf-sup condition, i.e.,
\begin{equation} \beta\left\|e_{p}^{n+1}\right\| \leq \frac{\left(\nabla e_{p}^{n+1}, v\right)}{\|\nabla v\|}. \end{equation} | (3.62) |
Then, we can constrain the products of the right-hand side of (3.61) with an arbitrary v \in V as follows:
\begin{equation*} \begin{aligned} \frac{1}{\tau}\left(\rho^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right), v\right) &\leq C \tau^{-1}\left\|\sigma^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right)\right\| \| \nabla v \|, \\ -\eta\left(\Delta e_{u}^{n+1}, v\right)-\left(R_{u}^{n+1}, v\right) &\leq C\left(\tau^{2}+\left\|\nabla e_{u}^{n+1}\right\|\right) \| \nabla v \|, \end{aligned} \end{equation*} |
and
\begin{equation} \begin{split} &- \lambda\left(w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-w^{n+1} \nabla \phi^{n}, v\right)\\ = & -\lambda\left(w\left(t_{n+1}\right)\left(\nabla \phi\left(t_{n+1}\right)-\nabla \phi\left(t_{n}\right)\right)+w\left(t_{n+1}\right) \nabla e_{\phi}^{n}+\nabla \phi^{n} e_{w}^{n+1}, v\right)\\ \leq& \lambda \left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\| \int_{t_{n}}^{t_{n+1}} \nabla \phi_{t}(t) d t\right\|\| v\|+ \lambda\left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n}\right\|\left\|v\right\|\\ &+ \lambda\left\|\nabla \phi^{n}\right\|_{L^2}\left\|e_w^{n+1}\right\|_{L^{3}}\left\|v\right\|_{L^{6}}\\ \leq& C\left( \tau+\left\|\nabla e_{\phi}^{n}\right\|+\left\|\nabla e_{w}^{n+1}\right\|\right)\left\| \nabla v \right\|.\\ \end{split} \end{equation} | (3.63) |
For the second term on the right-hand side, using \|v\|_{L^{3}} \leq\|v\|^{\frac{1}{2}}\|\nabla v\|^{\frac{1}{2}} , we have
\begin{equation} \begin{aligned} &\left(\frac{\rho^{n+1}-\rho^{n}}{\tau}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right), v\right) \\ &\leq C \tau^{-1}\left\|\rho^{n+1}-\rho^{n}\right\|\left\|e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right\|_{L^{3}}\|v\|_{L^{6}} \\ &\leq C \tau^{-1}\left\|\rho^{n+1}-\rho^{n}\right\|\left\|e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right\|^{\frac{1}{2}}\left\|\nabla\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\|\nabla v\| \\ &\leq C\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\left\|\nabla\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\|\nabla v\|. \end{aligned} \end{equation} | (3.64) |
By the split method, we arrive at
\begin{equation} \begin{aligned} &\left(\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \widetilde{\boldsymbol{u}}^{n+1}, v\right) \\ & = \left(\rho^{n+1}\left(\left(\boldsymbol{u}\left(t_{n+1}\right)-\boldsymbol{u}\left(t_{n}\right)\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right)+\left(\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde{e}_{u}^{n+1}, v\right) \\ &\quad+\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right)+\left(e_{\rho}^{n+1}\left(\boldsymbol{u}\left(t_{n+1}\right) \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right). \end{aligned} \end{equation} | (3.65) |
Utilizing H \ddot{o} lder's inequality and Young's inequality, we derive
\begin{equation} \begin{split} &\left(\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \widetilde{\boldsymbol{u}}^{n+1}, v\right) \\ \leq& C\left(\tau+\left\| e_{\rho}^{n+1}\right\|+\left\|\nabla e_{u}^{n}\right\|+\left\|\nabla \tilde{e}_{u}^{n+1}\right\|\right) \| \nabla v\|.\\ \end{split} \end{equation} | (3.66) |
By adding up the inequalities and incorporating (3.62), we get the desired result.
In this section, we will present some numerical experiments to prove the validity and accuracy of our method. In the following simulation, for phase field \phi , chemical potential w and pressure p , we take the P1 finite element space (continuous piecewise linear), and for fluid velocity \boldsymbol{u} , we take the P2 finite element space. All experiments were conducted in Freefem++. We fixed \eta = 0.8, \lambda = 0.7, \gamma = 0.0006, \varepsilon = 0.1, T = 0.1, \rho_1 = 1 and \rho_2 = 3. The computational domain and the initial conditions were taken as
\begin{equation*} \begin{split} &\Omega = \{(x, y)\in R^2:x^2+y^2 < 1\}, \\ &\phi_0 = cos(\pi x)cos(\pi y), \\ &\boldsymbol{u}_0 = ( \pi cos(\pi x)sin(\pi y), -\pi sin(\pi x)cos(\pi y)).\\ \end{split} \end{equation*} |
Tables 1 and 2 verify that (3.4) is the first-order convergence rate O(\tau) of (\phi, \sigma\boldsymbol{u}, \boldsymbol{u}, \rho) , which is consistent with the conclusion obtained from theoretical analysis. It is only proved in Theorem 3.2 that the pressure is the half-order convergence rate O(\tau^{\frac{1}{2}}) because of technical reasons. However, the numerical results on p in Table 1 still reach the first-order optimal convergence rate O(\tau) . Tables 3 and 4 show the convergence rate with another set of parameters.
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |
Figure 1 shows the evolution of the total energy at \tau = 0.02 . The downward trend of the energy curve confirms that our scheme is unconditionally energy stable. We also see a downward trend in energy when using different parameters. The energy curves for different time steps are shown in Figure 2 as a result of keeping the other parameters unchanged. It can been found that the curves are very similar, which means that the scheme is robust against different time steps.
To solve the Cahn-Hilliard phase field model for two-phase incompressible flows with variable density, we have designed a novel time marching scheme, which can significantly improve the calculation efficiency. The method is efficient because we decoupled the pressure from the velocity and phase field. We have also proved the unconditional energy stability, presented the error analysis and provided various numerical examples to demonstrate the stability and accuracy of the scheme. In addition, the decoupling method developed in this paper is universally applicable, and this method is always applicable for the generation of an effective fully decoupling scheme.
The authors declare they have not used Artificial Intelligence tools (AI) in the creation of this article.
This work was supported by the Research Project Supported by Shanxi Scholarship Council of China (No. 2021-029), Shanxi Provincial International Cooperation Base and Platform Project (202104041101019), Shanxi Province Natural Science Research (202203021211129), Shanxi Province Natural Science Research (No. 202203021212249) and Special/Youth Foundation of Taiyuan University of Technology (No. 2022QN101).
All authors declare that they have no competing interests in this paper.
[1] |
M. V. Solodov, B. F. Svaiter, A new projection method for variational inequality problems, SIAM J. Control Optim., 37 (1999), 765-776. doi: 10.1137/S0363012997317475
![]() |
[2] |
Y. Xiao, H. Zhu, A conjugate gradient method to solve convex constrained monotone equations with applications in compressive sensing, J. Math. Anal. Appl., 405 (2013), 310-319. doi: 10.1016/j.jmaa.2013.04.017
![]() |
[3] |
J. Liu, S. Li, A projection method for convex constrained monotone nonlinear equations with applications, Comput. Math. Appl., 70 (2015), 2442-2453. doi: 10.1016/j.camwa.2015.09.014
![]() |
[4] | J. Liu, Y. Feng, A derivative-free iterative method for nonlinear monotone equations with convex constraints, Numer. Algorithms, 82 (2019) 245-262. |
[5] | A. H. Ibrahim, A. I. Garba, H. Usman, J. Abubakar, A. B. Abubakar, Derivative-free projection algorithm for nonlinear equations with convex constraints, Thai Journal of Mathematics, 18 (2020), 212-232. |
[6] | P. Kaelo, M. Koorapetse, A globally convergent projection method for a system of nonlinear monotone equations, Int. J. Comput. Math., 2020. |
[7] | A. B. Abubakar, J. Rilwan, S. E. Yimer, A. H. Ibrahim, I. Ahmed, Spectral three-term conjugate descent method for solving nonlinear monotone equations with convex constraints, Thai Journal of Mathematics, 18 (2020) 501-517. |
[8] |
A. H. Ibrahim, P. Kumam, W. Kumam, A Family of Derivative-Free Conjugate Gradient Methods for Constrained Nonlinear Equations and Image Restoration, IEEE Access, 8 (2020), 162714- 162729. doi: 10.1109/ACCESS.2020.3020969
![]() |
[9] | A. B. Abubakar, A. H. Ibrahim, A. B. Muhammad, C. Tammer, A modified descent Dai-Yuan conjugate gradient method for constraint nonlinear monotone operator equations, Applied Analysis and Optimization, 4 (2020), 1-24. |
[10] | K. Meintjes, A. P. Morgan, A methodology for solving chemical equilibrium systems, Appl. Math. Comput., 22 (1987), 333-361. |
[11] |
K. Meintjes, A. P. Morgan, Chemical equilibrium systems as numerical test problems, ACM T. Math. Software, 16 (1990), 143-151. doi: 10.1145/78928.78930
![]() |
[12] |
M. W. Berry, M. Browne, A. N. Langville, V. P. Pauca, R. J. Plemmons, Algorithms and applications for approximate nonnegative matrix factorization, Comput. stat. data an., 52 (2007), 155-173. doi: 10.1016/j.csda.2006.11.006
![]() |
[13] |
S. P. Dirkse, M. C. Ferris, Mcplib: A collection of nonlinear mixed complementarity problems, Optim. method. softw., 5 (1995), 319-345. doi: 10.1080/10556789508805619
![]() |
[14] | J. E. Dennis, J. J. Moré, A characterization of superlinear convergence and its application to quasi-Newton methods, Math. comput., 28 (1974), 549-560. |
[15] |
R. Fletcher, C. M. Reeves, Function minimization by conjugate gradients, Comput. J., 7 (1964), 149-154. doi: 10.1093/comjnl/7.2.149
![]() |
[16] | R. Fletcher, Practical Methods of Optimization, John Wiley & Sons, 2013. |
[17] |
Y. H. Dai, Y. Yuan, A nonlinear conjugate gradient method with a strong global convergence property, SIAM J. optimiz., 10 (1999), 177-182. doi: 10.1137/S1052623497318992
![]() |
[18] |
M. R. Hestenes, E. Stiefel, Methods of conjugate gradients for solving linear systems, Journal of research of the National Bureau of Standards, 49 (1952), 409-436. doi: 10.6028/jres.049.044
![]() |
[19] | E. Polak, G. Ribiere, Note sur la convergence de méthodes de directions conjuguées, ESAIM-Math. Model. Num., 3 (1969), 35-43. |
[20] |
B. T. Polyak, The conjugate gradient method in extremal problems, U. S. S. R. Comput. Math. Math. Phys., 9 (1969), 94-112. doi: 10.1016/0041-5553(69)90035-4
![]() |
[21] | Y. Liu, C. Storey, Efficient generalized conjugate gradient algorithms, part 1: theory, J. optimiz. theory app., 69 (1991), 129-137. |
[22] | S. S. Djordjevi?, New Hybrid Conjugate Gradient Method As A Convex Combination of Ls and Fr Methods, Acta Math. Sci., 39 (2019), 214-228. |
[23] | A. H. Ibrahim, P. Kumam, A. B. Abubakar, W. Jirakitpuwapat, J. Abubakar, A hybrid conjugate gradient algorithm for constrained monotone equations with application in compressive sensing, Heliyon, 6 (2020), 1-17. |
[24] |
D. H. Li, M. Fukushima, A modified BFGS method and its global convergence in nonconvex minimization, J. Comput. Appl. Math., 129 (2001), 15-35. doi: 10.1016/S0377-0427(00)00540-9
![]() |
[25] |
D. H. Li, M. Fukushima, On the global convergence of the BFGS method for nonconvex unconstrained optimization problems, SIAM J. Optimiz., 11 (2001), 1054-1064. doi: 10.1137/S1052623499354242
![]() |
[26] |
E. G. Birgin, J. M. Martinez, A spectral conjugate gradient method for unconstrained optimization, Appl. Math. opt., 43 (2001), 117-128. doi: 10.1007/s00245-001-0003-0
![]() |
[27] |
G. Yuan, T. Li, W. Hu, A conjugate gradient algorithm for large-scale nonlinear equations and image restoration problems, Appl. Numer. Math., 147 (2020), 129-141. doi: 10.1016/j.apnum.2019.08.022
![]() |
[28] | D. H. Li, X. L. Wang, A modified Fletcher-Reeves-type derivative-free method for symmetric nonlinear equations, Numerical Algebra, Control & Optimization, 1 (2011), 71-82. |
[29] | E. D. Dolan, J. J. Moré, Benchmarking optimization software with performance profiles, Math. program., 91 (2002), 201-213. |
[30] | A. B. Abubakar, P. Kumam, H. Mohammad, A. M. Awwal, K, Sitthithakerngkiet, A modified fletcher-reeves conjugate gradient method for monotone nonlinear equations with some applications, Mathematics, 7 (2019), 1-25. |
[31] |
W. La Cruz, J. Martínez, M. Raydan, Spectral residual method without gradient information for solving large-scale nonlinear systems of equations, Math. Comput., 75 (2006), 1429-1448. doi: 10.1090/S0025-5718-06-01840-0
![]() |
[32] | W. La Cruz, A spectral algorithm for large-scale systems of nonlinear monotone equations, Numer. Algorithms, 76 (2017), 1109-1130. |
[33] |
Y. Bing, G. Lin, An efficient implementation of Merrill's method for sparse or partially separable systems of nonlinear equations, SIAM J. Optimiz., 1 (1991), 206-221. doi: 10.1137/0801015
![]() |
[34] |
Z. Yu, J. Lin, J. Sun, Y. Xiao, L. Liu, Z. Li, Spectral gradient projection method for monotone nonlinear equations with convex constraints, Appl. Numer. Math., 59 (2009), 2416-2423. doi: 10.1016/j.apnum.2009.04.004
![]() |
[35] |
Y. Ding, Y. Xiao, J. Li, A class of conjugate gradient methods for convex constrained monotone equations, Optimization, 66 (2017), 2309-2328. doi: 10.1080/02331934.2017.1372438
![]() |
[36] | I. Daubechies, M. Defrise, C. De Mol, An iterative thresholding algorithm for linear inverse problems with a sparsity constraint, Commun. Pur. Appl. Math., 57 (2004), 1413-1457. |
[37] |
A. Beck, M. Teboulle, A fast iterative shrinkage-thresholding algorithm for linear inverse problems, SIAM J. Imaging Sci., 2 (2009), 183-202. doi: 10.1137/080716542
![]() |
[38] | E. T. Hale, W. Yin, Y. Zhang, A fixed-point continuation method for l1-regularized minimization with applications to compressed sensing, CAAM TR07-07, Rice University, 2007, 1-45. |
[39] |
S. Huang, Z. Wan, A new nonmonotone spectral residual method for nonsmooth nonlinear equations, J. Comput. Appl. Math., 313 (2017), 82-101. doi: 10.1016/j.cam.2016.09.014
![]() |
[40] | J. Abubakar, P. Kumam, A. H. Ibrahim, A. Padcharoen, Relaxed Inertial Tseng's Type Method for Solving the Inclusion Problem with Application to Image Restoration, Mathematics, 8 (2020), 1-19. |
[41] |
A. H. Ibrahim, P. Kumam, A. B. Abubakar, J. Abubakar, A. B. Muhammad, Least-Square-Based Three-Term Conjugate Gradient Projection Method for?1-Norm Problems with Application to Compressed Sensing, Mathematics, 8 (2020), 1-21. doi: 10.3390/math8101793
![]() |
[42] | A. B. Abubakar, K. MUangchoo, A. Muhammad, A. H. Ibrahim, A Spectral Gradient Projection Method for Sparse Signal Reconstruction in Compressive Sensing, Modern Applied Science, 14 (2020), 86-93. |
[43] | A. C. Bovik, Handbook of Image and Video Processing, Academic press, 2010. |
[44] |
S. M. Lajevardi, Structural similarity classifier for facial expression recognition, Signal Image Video P., 8 (2014), 1103-1110. doi: 10.1007/s11760-014-0639-2
![]() |
1. | Yuxia Guo, Yichen Hu, Infinitely many solutions for Hamiltonian system with critical growth, 2024, 13, 2191-950X, 10.1515/anona-2023-0134 | |
2. | Xingyue He, Chenghua Gao, Jingjing Wang, k-convex solutions for multiparameter Dirichlet systems with k-Hessian operator and Lane-Emden type nonlinearities, 2024, 13, 2191-950X, 10.1515/anona-2023-0136 | |
3. | Hongying Jiao, Shuhai Zhu, Jinguo Zhang, Existence of infinitely many solutions for critical sub-elliptic systems via genus theory, 2024, 16, 2836-3310, 237, 10.3934/cam.2024011 | |
4. | Jinli Yang, Jiajing Miao, Algebraic Schouten solitons of Lorentzian Lie groups with Yano connections, 2023, 15, 2836-3310, 763, 10.3934/cam.2023037 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |