In this paper, we consider the mixed finite element methods for quadratic optimal control problems governed by convective diffusion equations. The state and the co-state are discretized by the lowest order Raviart-Thomas mixed finite element spaces and the control is approximated by piecewise constant functions. Using some proper duality problems, we derive a posteriori\(L^{2}(0,T;L^{2}(\Omega))\) error estimates for the scalar functions. Such estimates, which are apparently not available in the literature, are an important step toward developing reliable adaptive mixed finite element approximation schemes for the control problem.
Hinweise
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
The first author carried out the molecular genetic studies, participated in the sequence alignment, and drafted the manuscript. The second author conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.
1 Introduction
As far as we know, optimal control problems [1] have been extensively utilized in many aspects of the modern life such as social, economic, scientific, and engineering numerical simulation. Thus, they must be solved successfully with efficient numerical methods. Among these numerical methods, finite element method is a good choice. There have been extensive studies in the convergence of finite element approximation of optimal control problems; see [2‐6]. A systematic introduction to finite element methods for PDEs and optimal control can be found for example in [7‐9].
Recently, the adaptive finite element method has been investigated extensively. It has become one of the most popular methods in the scientific computation and numerical modeling. An adaptive finite element approximation ensures a higher density of nodes in a certain area of the given domain, where the solution is more difficult to approximate, indicated by a posteriori error estimators. Hence it is an important approach to boost the accuracy and efficiency of finite element discretizations. There are lots of works concentrating on the adaptivity of various optimal control problems. See, for example, [10‐19].
Anzeige
In many control problems, the objective functional contains the gradient of the state variables. Thus, the accuracy of the gradient is important in numerical discretization of the coupled state equations. Mixed finite element methods are appropriate for the state equations in such cases since both the scalar variable and its flux variable can be approximated to the same accuracy by using such methods; see, for example, [20‐23].
We shall use the lowest order Raviart-Thomas mixed finite element to discretize the state and the co-state, and use the piecewise constant space to approximate the control variable. Using some proper duality problems, we derive a posteriori\(L^{2}(0,T;L^{2}(\Omega))\) error estimates for the scalar functions. The optimal control problems that we are interested in are as follows:
where the bounded open set \(\Omega\subset{\mathbf{R}^{2}}\) is a convex polygon with the boundary ∂Ω. \(J=[0,T]\). Let K be a closed convex set in the control space \(U=L^{2}(J;L^{2}(\Omega ))\), \(\mathbf{p}, \mathbf{p}_{d}\in(L^{2}(J;H^{1}(\Omega)))^{2}\), \(u, y, y_{d}\in L^{2}(J;H^{1}(\Omega))\), \(f\in L^{2}(J;L^{2}(\Omega))\), \(y_{0}(x)\in H_{0}^{1}(\Omega)\). Moreover, we assume that \(0< a_{0}\leq a\leq a^{0}\), \(a(x)\in W^{1,\infty}(\Omega)\), \(c(x)\in W^{1,\infty}(\Omega)\), \(\mathbf{b}(x)\in(W^{1,\infty}(\Omega))^{2}\).
We assume that the constraint on the control is an obstacle such that
In this paper, we adopt the standard notation \(W^{m,p}(\Omega)\) for Sobolev spaces on Ω with a norm \(\|\cdot\|_{m,p}\) given by \(\| v \|_{m,p}^{p}=\sum_{|\alpha|\leq m}\| D^{\alpha}v\|_{L^{p}(\Omega)}^{p}\), a semi-norm \(|\cdot|_{m,p}\) given by \(| v|_{m,p}^{p}=\sum_{|\alpha|= m}\| D^{\alpha}v\|_{L^{p}(\Omega)}^{p}\). We set \(W_{0}^{m,p}(\Omega)=\{v\in W^{m,p}(\Omega): v|_{\partial\Omega}=0\}\). For \(p=2\), we denote \(H^{m}(\Omega)=W^{m,2}(\Omega)\), \(H_{0}^{m}(\Omega)=W_{0}^{m,2}(\Omega)\), and \(\|\cdot\|_{m}=\|\cdot\|_{m,2}\), \(\|\cdot\|=\|\cdot\|_{0,2}\).
Anzeige
We denote by \(L^{s}(0,T;W^{m,p}(\Omega))\) the Banach space of all \(L^{s}\) integrable functions from J into \(W^{m,p}(\Omega)\) with norm \(\| v\|_{L^{s}(J;W^{m,p}(\Omega))}= (\int_{0}^{T}\|v\|_{W^{m,p}(\Omega )}^{s}\, dt )^{\frac{1}{s}}\) for \(s\in[1,\infty)\), and the standard modification for \(s=\infty\). Similarly, one can define the spaces \(H^{l}(J;W^{m,p}(\Omega))\) and \(C^{k}(J;W^{m,p}(\Omega))\). In addition C denotes a general positive constant independent of h and Δt, where h is the spatial mesh-size for the control and state discretization and Δt is the time increment.
The plan of this paper is as follows. In next section, we shall give a brief review on the mixed finite element method and the backward Euler discretization, and then we construct the approximation for the optimal control problems (1.1)-(1.4). Then, using two duality problems, we derive a posteriori\(L^{2}(0,T;L^{2}(\Omega))\) error estimates for the scalar functions in Section 3. Finally, we give a conclusion and indicate some possible future work.
2 Mixed methods of parabolic optimal control problems
In this section, we shall study the mixed finite element and the backward Euler discretization approximation of convective diffusion optimal control problems (1.1)-(1.4). For the sake of simplicity, we assume that the domain Ω is a convex polygon. Now, we introduce the co-state parabolic equation
Let \(\alpha=a^{-1}\) and \(\boldsymbol{\beta}=\alpha\mathbf{b}\). We recast (1.1)-(1.4) as the following weak form: find \((\mathbf{p},y,u)\in\mathbf{L}\times Q\times K\) such that
It follows from [1] and [16] that the optimal control problem (2.4)-(2.7) has a unique solution \((\mathbf{p},y,u)\), and that a triplet \((\mathbf{p},y,u)\) is the solution of (2.4)-(2.7) if and only if there is a co-state \((\mathbf{q},z)\in\mathbf{L}\times Q\) such that \(({\mathbf {p}},y,{\mathbf{q}},z,u)\) satisfies the following optimality conditions:
where \((\cdot,\cdot)\) is the inner product of \(L^{2}(\Omega)\).
Let \({\mathcal{T}}_{h}\) be regular triangulations of Ω. \(h_{\tau}\) is the diameter of τ and \(h=\max h_{\tau}\). Let \(\mathbf{V}_{h}\times W_{h}\subset\mathbf{V}\times W\) denote the lowest order Raviart-Thomas space [24] associated with the triangulations \({\mathcal{T}}_{h}\) of Ω. \(P_{k}\) denotes the space of polynomials of total degree of at most k (\(k\geq0\)). Let \(\mathbf{V}({\tau})=\{\mathbf{v}\in P_{0}^{2}({\tau})+x\cdot P_{0}({\tau })\}\), \(W({\tau})=P_{0}({\tau})\). We define
The mixed finite element discretization of (2.4)-(2.7) is as follows: compute \((\mathbf{p}_{h},y_{h},u_{h})\in L^{2}(J;\mathbf {V}_{h})\times H^{1}(J;W_{h})\times K_{h}\) such that
where \(y_{0}^{h}(x)\in W_{h}\) is an approximation of \(y_{0}\). The optimal control problem (2.15)-(2.18) again has a unique solution \((\mathbf{p}_{h},y_{h},u_{h})\), and that a triplet \(({\mathbf {p}}_{h},y_{h},u_{h})\) is the solution of (2.15)-(2.18) if and only if there is a co-state \((\mathbf{q}_{h},z_{h})\in L^{2}(J;\mathbf{V}_{h})\times H^{1}(J;W_{h})\) such that \((\mathbf{p}_{h},y_{h},\mathbf{q}_{h},z_{h},u_{h})\) satisfies the following optimality conditions:
We now consider the fully discrete approximation for the above semidiscrete problem. Let \(\Delta t>0\), \(N=T/\Delta t\in \mathbb{Z}\), and \(t_{i}=i \Delta t\), \(i\in\mathbb{Z}\). Also, let
We address the fully discrete approximation scheme to find \((\mathbf{p}_{h}^{i},y_{h}^{i},u_{h}^{i})\in\mathbf{V}_{h}\times W_{h}\times K_{h}\), \(i=1, 2, \ldots, N\), such that
where \(f^{i}=f^{i}(x)=f(x,t_{i})\), \(y_{d}^{i}=y_{d}(x,t_{i})\), and \(\mathbf {p}_{d}^{i}=\mathbf{p}_{d}(x,t_{i})\).
It follows that the control problem (2.26)-(2.29) has a unique solution \((\mathbf{p}_{h}^{i},y_{h}^{i},u_{h}^{i})\), \(i=1, 2, \ldots, N\), and that a triplet \((\mathbf{p}_{h}^{i},y_{h}^{i},u_{h}^{i})\in\mathbf {V}_{h}\times W_{h}\times K_{h}\), \(i=1, 2, \ldots, N\), is the solution of (2.26)-(2.29) if and only if there is a co-state \(({\mathbf{q}}_{h}^{i-1},z_{h}^{i-1})\in\mathbf{V}_{h}\times W_{h}\) such that \(({\mathbf{p}}_{h}^{i},y_{h}^{i},{\mathbf{q}}_{h}^{i-1},z_{h}^{i-1},u_{h}^{i})\in (\mathbf{V}_{h}\times W_{h})^{2}\times K_{h}\) satisfies the following optimality conditions:
In the rest of the paper, we shall use some intermediate variables. For any control function \(U_{h}\in K_{h}\), we first define the state solution \((\mathbf{p}(U_{h}),y(U_{h}),\mathbf{q}(U_{h}),z(U_{h}))\) to satisfy
Let \(\Pi_{h}:\mathbf{V}\rightarrow\mathbf{V}_{h}\) be the Raviart-Thomas projection operator [26], which satisfies: for any \(\mathbf{v}\in\mathbf{V}\),
Firstly, let us derive the a posteriori error estimates for the control u.
Theorem 3.1
Let\((y,\mathbf{p},z,\mathbf{q},u)\)and\((Y_{h},P_{h},Z_{h},Q_{h},U_{h})\)be the solutions of (2.8)-(2.14) and (2.39)-(2.45), respectively. Then we have
$$ \|u-U_{h}\|_{L^{2}(J;L^{2}(\Omega))}^{2}\leq C \eta_{1}^{2} + \bigl\Vert \tilde{Z}_{h}-z(U_{h}) \bigr\Vert _{L^{2}(J;L^{2}(\Omega))}^{2}, $$
In order to estimate the error \(\|\tilde{Z}_{h}-z(U_{h})\|_{L^{2}(J;L^{2}(\Omega))}^{2}\), we need the following well-known stability results (see [27, 28] for the details) for the following dual equations:
Letfandgbe piecewise continuous nonnegative functions defined on\(0\leq t \leq T\), gbeing non-decreasing. If, for each\(t\in J\),
$$ f(t)\leq g(t)+\int_{0}^{t}f(s)\, d s, $$
(3.12)
then\(f(t)\leq e^{t}g(t)\).
In the following two theorems, we shall estimate the error \(\|\tilde {Z}_{h}-z(U_{h})\|_{L^{2}(J;L^{2}(\Omega))}\).
Theorem 3.2
Let\((Y_{h},P_{h},Z_{h},Q_{h},U_{h})\)and\((y(U_{h}),\mathbf{p}(U_{h}),z(U_{h}),\mathbf {q}(U_{h}),U_{h})\)be the solutions of (2.39)-(2.45) and (2.46)-(2.51), respectively. Then we have
Let\((y,\mathbf{p},z,\mathbf{q},u)\)and\((Y_{h},P_{h},Z_{h},Q_{h},U_{h})\)be the solutions of (2.8)-(2.14) and (2.39)-(2.45), respectively. Let\((y(U_{h}),\mathbf{p} (U_{h}),z(U_{h}),\mathbf{q}(U_{h}),U_{h})\)be defined as in (2.46)-(2.51). Then we have the following error estimate:
The triangle inequality and (3.30) yield (3.21). □
Remark 3.1
If we use the higher order RT mixed finite elements to approximate the state variables and the co-state variables, then the estimators \(\eta _{2}^{2}\), \(\eta_{3}^{2}\), \(\eta_{8}^{2}\), and \(\eta_{9}^{2}\) in Theorem 3.2 and Theorem 3.3 can be improved by
where \(\nabla_{h} \chi|_{\tau}=\nabla(\chi|_{\tau})\).
Let \(({\mathbf{p}},y,{\mathbf{q}},z,u)\) and \((P_{h},Y_{h},Q_{h},Z_{h},U_{h})\) be the solutions of (2.8)-(2.14) and (2.39)-(2.45), respectively. We decompose the errors as follows:
Using (3.41) and (3.35), we complete the proof of Theorem 3.4. □
Collecting Theorems 3.1-3.4, we can derive the following results.
Theorem 3.5
Let\((\mathbf{p},y,\mathbf{q},z,u)\)and\((P_{h},Y_{h}, Q_{h},Z_{h},U_{h})\)be the solutions of (2.8)-(2.14) and (2.39)-(2.45), respectively. Then we have
where\(\eta_{1}\)is defined in Theorem 3.1, \(\eta_{2},\ldots,\eta_{7}\)are defined in Theorem 3.2, and\(\eta_{8},\ldots,\eta_{14}\)are defined in Theorems3.3, respectively.
Acknowledgements
The first author is supported by the scientific research program in Hunan University of Science and Engineering (2015). The second author is supported by the National Natural Science Foundation of China (11401201), the Foundation of Hunan Educational Committee (13C338), and the construct program of the key discipline in Hunan University of Science and Engineering.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
The first author carried out the molecular genetic studies, participated in the sequence alignment, and drafted the manuscript. The second author conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.