Table of Contents Author Guidelines Submit a Manuscript
Advances in Mathematical Physics
Volume 2017 (2017), Article ID 6801260, 6 pages
https://doi.org/10.1155/2017/6801260
Research Article

Conditional Well-Posedness for an Inverse Source Problem in the Diffusion Equation Using the Variational Adjoint Method

1School of Science, Shandong University of Technology, Zibo 255049, China
2Department of Mathematics, Southeast University, Nanjing 210096, China

Correspondence should be addressed to Gongsheng Li

Received 22 March 2017; Accepted 31 May 2017; Published 27 June 2017

Academic Editor: Mikhail Panfilov

Copyright © 2017 Chunlong Sun et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

This article deals with an inverse problem of determining a linear source term in the multidimensional diffusion equation using the variational adjoint method. A variational identity connecting the known data with the unknown is established based on an adjoint problem, and a conditional uniqueness for the inverse source problem is proved by the approximate controllability to the adjoint problem under the condition that the unknowns can keep orders locally. Furthermore, a bilinear form is set forth also based on the variational identity and then a norm for the unknowns is well-defined by which a conditional Lipschitz stability is established.

1. Introduction

Air pollution and fog-haze problems have attracted much attention during the last decade in North China. There are some important aspects in the research of air pollution, such as the component and property of the air-pollutant, the accumulating process, and the migrating rule and the diffusion model. It is an effective method to analyze the transport and diffusion behaviors utilizing suitable advection-diffusion equations (cf. [13], for instance). However, for real-life diffusion problems, some physical parameters are always unknown, or can not be measured directly, or can be obtained but expanding much cost, such as the diffusion coefficient, the initial distribution of the pollutant, and the physical/chemical source term. Therefore, it becomes feasible and necessary to put forward inverse problems using some additional data to determine those parameters with less cost in mathematics. An inverse problem arising in a diffusion equation belongs to inverse problems of parabolic type of partial differential equations.

By the literatures we have, there are quite a few of studies on inverse problems for parabolic equations since 1970s. As for general researches and summary, we refer to the monograph [4], and, for the research methods, we refer to [515] for the fixed point method based on the solution’s expression of the forward problem, refer to [1619] for the orthogonality method and energy estimates method, refer to [2023] for utilizing the maximum principle, and refer to [2428] for Carleman-type estimates method, and so forth. It is noted that stability analysis is still a trouble for inverse problems, especially for the construction of Lipschitz stability. By using classical estimates for parabolic problems in Hölder spaces, Hölder stability can be obtained based on the maximum principles and Sobolev embedding theorems; see also [4, 21, 22] and so forth. However, such method always needs more conditions for the initial boundary value functions, and it always involves complicated integral estimates. On the other hand, the variational identity method, also known as monotonicity method or the adjoint method, see, for example, [2935], has been applied to parameter identification problems in the parabolic equations, by which uniqueness results can be proved using approximate controllability for the adjoint problems based on integral identities. The author discussed some inverse source problems for parabolic equations in 1D case (see [3638] for instance) and gave conditional stability estimates for determining the source term or source coefficient using the variational identity method. Recently, an inverse problem of determining the first-order coefficient in an advection-dispersion equation in 2D/3D case using the final observations was discussed also using the variational adjoint method [39], but the research domain is limited to be regular and the unknown should keep a form of variables separable, and the solution to the adjoint problem should have an explicit expression in order to prove the Lipschitz stability of the inverse problem.

In this paper we continue to deal with the inverse source problem for the multidimensional diffusion equation in a bounded open domain in . Here the inverse source problem is to determine the space-dependent source coefficient using the flux at partial boundary. A variational identity connecting the known data with the unknown source coefficient is established based on a suitable adjoint problem, and the uniqueness of the inverse problem is proved by the approximate controllability to the adjoint problem under the condition that the unknowns can keep orders locally. Furthermore, a bilinear form is set forth by the variational identity and then a norm for the unknowns is well-defined by which a conditional Lipschitz stability is constructed. It is noted that the maximum-minimum principle for parabolic equations is employed to keep sign of the solution to the adjoint problem instead of utilizing the explicit expression of the solution to the adjoint problem in the proof of the stability. Such kind of research approach can give more wide applications of the variational adjoint method to study well-posedness of inverse problems arising in partial differential equations.

2. The Forward Problem and the Inverse Source Problem

2.1. The Forward Problem

Let and , denote , and let be the smooth boundary of . Consider the multidimensional diffusion equation:where denotes the concentration of the pollutant at the space point and the time , is the Laplace operator, is the diffusion coefficient, and is a linear source term. Generally speaking, the source term has a variables separable form: , where is a time-dependent attenuation factor and is a space-dependent source magnitude.

The initial boundary value conditions are given as follows:

If the model parameters and the initial boundary value functions in the forward problem (1)-(2) are all known and satisfy suitable consistency conditions, then it is a well-posed deterministic problem by the theory of parabolic equations, and there exists a unique solution on . Here we omit some related assertions on the forward problem (1)-(2) and focus on the inverse source problem given in the following.

2.2. The Inverse Source Problem

Suppose that there occurs some polluting phenomenon in a bounded region, and the pollutant is mainly produced by some continuous source distributed in the region. If the source term can not be measured directly, an inverse problem is encountered. Let the time-dependent factor in the source term be known; we are to determine the space-dependent source magnitude function using the additional flux data measured at the partial boundary of . That is, we have the additional condition:where denotes the normal outward vector at the boundary . Thus an inverse source problem is formulated by the problem (1)-(2) together with the overposed information (3).

Denote , and let be an admissible set for the unknowns, where is a positive constant. For any given , there is a unique solution to the forward problem, denoted by . Together with the additional information, we define a mapping : Then the inverse problem (1)–(3) can be transformed to the problem of solving the operator equation (4). From the viewpoint of numerics, solving the inverse problem is reduced to solve the minimization problem with regularization strategywhere is the regularization parameter. This paper is devoted to the well-posedness of the inverse source problem; numerical inversions will be presented in another occasion. Therefore, the following lemma on the approximate controllability to the heat equation is useful.

Lemma 1 (see [40]). Let be a bounded open subset with piecewise-smooth boundary having finite -dimensional volume, be the set of all admissible boundary functions , and be any nonempty set of . For , let be the solution to the following initial boundary value problem:Then for any given and , there exists a function , continuous in , such that

The assertion of this lemma is also valid to general parabolic equations, and it reveals that the final value of a homogeneous linear parabolic model can be approximately controlled by its boundary value.

3. Uniqueness of the Inverse Source Problem

3.1. The Variational Identity

We give a variational identity connecting the known data with the unknown source function based on a suitable adjoint problem.

Theorem 2. Suppose that the initial and boundary value functions and are given and fixed here. Let and be the solutions to the forward problem (1)-(2) corresponding to and be the additional flux data measured at . Then there holdswhere satisfies the adjoint problemAnd is a controllable boundary input.

Proof. Denote ; we haveand the additional informationBy smooth test function , multiplying the two sides of the first equation in (10), and integrating on , there holdsNoting the homogeneous boundary conditions in (10), we have by integration by parts for the left-hand side of (12)Let be the solution to the adjoint problem (9); then it follows that (8) is valid by (13) together with the condition (11). Since the adjoint problem (9) is uniquely determined by its boundary input , we denote its solution as . The proof is over.

3.2. Conditional Uniqueness

Uniqueness is an important aspect in the research of inverse problems. With the help of the variational identity (8) and Lemma 1, we will prove a conditional uniqueness for the inverse source problem given above. The required condition is that the unknown function should keep its sign in a small region of the considered domain.

Theorem 3. Under the conditions of Theorem 2, also assume that and for . If on , then there is no positive measurable subdomain of where is of one sign.

Proof. By the assumption of on , we have by identity (8)Suppose that there is a subdomain with positive measure such that keeps one sign. Let for convenience. We will deduce a contradiction with (14) in the following.
DenoteTake a partition of the time interval :and denote , ; we havewhere denotes the truncated error and is a positive constant. Noting the condition , here the term of is omitted.
By the assumption, there exists a nonnegative integrable function on given bywhere is a smooth positive function on . Thus we getIn addition, we define a series of integrable functions on by the assumption on :Noting that the adjoint problem (9) is a well-posed backward problem and the initial state is zero at , we deduce that its solution at can be approximated to any objective function by Lemma 1 by suitably choosing the boundary input . Subsequently, there exists a boundary input on each subinterval such that for any we haveBased on (17) and (19), utilizing the basic inequality and noting , we getThanks to , there holds the estimate for the center term of the right-hand side in (22) by (21) and Cauchy-Schwartz inequalityHenceforth, we haveIt follows that there must be as long as and are small enough, which is a contradiction with (14). The proof is completed.

4. Conditional Lipschitz Stability

Firstly we give a basic assertion.

Lemma 4. Let be bounded measurable domain. Suppose that and for ; then implies that a.e. .

With a complete method as that used in Theorem 2, we can prove the following assertion.

Theorem 5. Let , , , and be the solutions of the forward problem (1)-(2) corresponding to , , and , , and be the additional flux data. Then there holdswhere is also the solution to the adjoint problem (9).

For construction of stability for the inverse source problem, we need extra conditions for the known and unknown data functions. For the unknown source coefficient function , we further assume that and for . In addition we assume that the attenuation factor is continuous and positive on and ; here is a positive constant. Now, based on the variational identity (8), we define a functional on the source function and the controllable input :where is the solution to the adjoint problem (9) which only depends upon the boundary input for .

It is not difficult to show that the bilinear functional is bounded on and in norm. Then we define a norm on which is called -norm given by

We need to prove that the norm given by (27) is well-defined. Obviously, is nonnegative and satisfies the absolute homogeneity and triangle inequality due to the linearity and additivity of the integration. We now prove that if , then there must be a.e. in the case of and .

In fact, assume that ; that is, there holdsBy the assumptions that for we deduce that the solution to the adjoint problem (9) takes nonnegative values by the maximum-minimum principle for the parabolic equations (cf. [41], for instance). Together with and it follows that the integrated function in (28) takes nonnegative values; that is, we getNoting the assumption on the attenuation factor , there must be a.e. or a.e. by Lemma 4. Since the boundary input satisfies the condition and , there holds a.e. also by the maximum-minimum principle. So there must be a.e. . Thus the norm given by (27) is well-defined.

Theorem 6. Under the conditions of Theorem 5, let and take nonnegative values on , and let be continuous and positive on and . Then there exists a positive constant such thatwhere the norm is defined by (27).

Proof. By the variational identity (25) and the norm definition (27), utilizing Cauchy-Schwartz inequality, we haveBy general regularity theory of parabolic equations with the initial boundary value conditions, we know that, for , there exists a constant such that Henceforth we getNoting (31) and the condition , by setting it follows that the assertion of this theorem is valid.

5. Conclusions

An inverse problem of determining a space-dependent source coefficient in the multidimensional diffusion equation using the variational adjoint method is investigated. A variational identity connecting the known data with the unknown is established based on an adjoint problem, and uniqueness for the inverse source problem is proved by the approximate controllability to the adjoint problem under suitable conditions. This is the conditional uniqueness. Also, based on the variational identity, a bilinear form is set forth by which a norm for the unknown source is well-defined, and then a Lipschitz stability can be proved if the unknown source is of one sign. Such method can be generalized to study the Lipschitz stability for other kinds of inverse coefficient problems arising in partial differential equations.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Acknowledgments

This work is partially supported by the National Natural Science Foundation of China (nos. 11371231, 11071148).

References

  1. G. S. Li and D. Yao, The inversion Problems of Source Terms and Applications, Science Press, Beijing, 2014.
  2. N.-Z. Sun, Mathematical modeling of groundwater pollution, Springer-Verlag, New York, 1996. View at Publisher · View at Google Scholar · View at MathSciNet
  3. C. M. Zheng and G. D. Bennett, “Applied Contaminant Transport Modeling,” John Wiley Sons, New York, NY, USA, 2002. View at Google Scholar
  4. V. Isakov, Inverse Problems for Partial Differential Equations, vol. 127 of Applied Mathematical Sciences, Springer, New York, NY, USA, 1998. View at Publisher · View at Google Scholar · View at MathSciNet
  5. J. R. Cannon, The One-Dimensional Heat Equation, Addison-Wesley, Reading, Mass, USA, 1984. View at Publisher · View at Google Scholar · View at MathSciNet
  6. J. R. Cannon and H. Yin, “A uniqueness theorem for a class of nonlinear parabolic inverse problems,” Inverse Problems, vol. 4, no. 2, pp. 411–416, 1988. View at Publisher · View at Google Scholar · View at Scopus
  7. J. R. Cannon and S. Perez-Esteva, “Uniqueness and stability of 3D heat sources,” Inverse Problems, vol. 7, no. 1, pp. 57–62, 1991. View at Publisher · View at Google Scholar · View at Scopus
  8. P. DuChateau and W. Rundell, “Unicity in an inverse problem for an unknown reaction term in a reaction-diffusion equation,” Journal of Differential Equations, vol. 59, no. 2, pp. 155–164, 1985. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  9. S. Gatti, “An existence result for an inverse problem for a quasilinear parabolic equation,” Inverse Problems, vol. 14, no. 1, pp. 53–65, 1998. View at Publisher · View at Google Scholar · View at MathSciNet
  10. L. Gongsheng, M. Yichen, and L. Kaitai, “An inverse parabolic problem for nonlinear source term with nonlinear boundary conditions,” Journal of Inverse and Ill-Posed Problems, vol. 11, no. 4, pp. 371–387, 2003. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  11. A. Nanda and P. C. Das, “Determination of the source term in the heat conduction equation,” Inverse Problems, vol. 12, no. 3, pp. 325–339, 1996. View at Publisher · View at Google Scholar · View at MathSciNet
  12. M. Pilant and W. Rundell, “Fixed point methods for a nonlinear parabolic inverse coefficient problem,” Communications in Partial Differential Equations, vol. 13, no. 4, pp. 469–493, 1988. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  13. M. Pilant and W. Rundell, “A method for identifying nonlinear terms in parabolic initial-boundary value problems,” Advances in Water Resources, vol. 14, no. 2, pp. 83–88, 1991. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  14. W. Rundell, “The determination of a parabolic equation from initial and final data,” Proceedings of the American Mathematical Society, vol. 99, no. 4, pp. 637–642, 1987. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  15. A. Zeghal, “Existence results for inverse problems associated with a nonlinear parabolic equation,” Journal of Mathematical Analysis and Applications, vol. 272, no. 1, pp. 240–248, 2002. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  16. I. Bushuyev, “Global uniqueness for inverse parabolic problems with final observation,” Inverse Problems, vol. 11, no. 4, article no. 001, pp. L11–L16, 1995. View at Publisher · View at Google Scholar · View at Scopus
  17. V. Isakov, Inverse Source Problems, vol. 34 of Mathematical Surveys and Monographs, American Mathematical Society, Providence, RI, USA, 1990. View at Publisher · View at Google Scholar · View at MathSciNet
  18. E. G. Savateev and R. Riganti, “Inverse problem for the nonlinear heat equation with the final overdetermination,” Mathematical and Computer Modelling, vol. 22, no. 1, pp. 29–43, 1995. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  19. M. Yamamoto, “Conditional stability in determination of force terms of heat equations in a rectangle,” Mathematical and Computer Modelling, vol. 18, no. 1, pp. 79–88, 1993. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  20. M. Choulli and M. Yamamoto, “Conditional stability in determining a heat source,” Journal of Inverse and Ill-Posed Problems, vol. 12, no. 3, pp. 233–243, 2004. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  21. V. Isakov, “Inverse parabolic problems with the final overdetermination,” Communications on Pure and Applied Mathematics, vol. 44, no. 2, pp. 185–209, 1991. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  22. V. Isakov, “Some inverse problems for the diffusion equation,” Inverse Problems, vol. 15, no. 1, pp. 3–10, 1999. View at Publisher · View at Google Scholar · View at MathSciNet
  23. W. Rundell, “Determination of an unknown nonhomogeneous term in a linear partial differential equation from overspecified boundary data,” Applicable Analysis, vol. 10, no. 3, pp. 231–242, 1980. View at Publisher · View at Google Scholar · View at MathSciNet
  24. V. Isakov, “Carleman type estimates in an anisotropic case and applications,” Journal of Differential Equations, vol. 105, no. 2, pp. 217–238, 1993. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  25. M. V. Klibanov, “Inverse problems and Carleman estimates,” Inverse Problems, vol. 8, no. 4, article no. 009, pp. 575–596, 1992. View at Publisher · View at Google Scholar · View at Scopus
  26. M. V. Klibanov and A. Timonov, Carleman estimates for coefficient inverse problems and numerical applications, Inverse and Ill-posed Problems Series, VSP, Utrecht, 2004. View at Publisher · View at Google Scholar · View at MathSciNet
  27. M. Machida and M. Yamamoto, “Global Lipschitz stability in determining coefficients of the radiative transport equation,” Inverse Problems, vol. 30, no. 3, Article ID 035010, 2014. View at Publisher · View at Google Scholar · View at MathSciNet
  28. M. Yamamoto, “Carleman estimates for parabolic equations and applications,” Inverse Problems, vol. 25, no. 12, Article ID 123013, 123013, 75 pages, 2009. View at Publisher · View at Google Scholar · View at MathSciNet
  29. J. R. Cannon and P. DuChateau, “Structural identification of an unknown source term in a heat equation,” Inverse Problems, vol. 14, no. 3, pp. 535–551, 1998. View at Publisher · View at Google Scholar · View at MathSciNet
  30. P. DuChateau, “Monotonicity and invertibility of coefficient-to-data mappings for parabolic inverse problems,” SIAM Journal on Mathematical Analysis, vol. 26, no. 6, pp. 1473–1487, 1995. View at Publisher · View at Google Scholar · View at MathSciNet
  31. P. DuChateau, “An inverse problem for the hydraulic properties of porous media,” SIAM Journal on Mathematical Analysis, vol. 28, no. 3, pp. 611–632, 1997. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  32. P. DuChateau, R. Thelwell, and G. Butters, “Analysis of an adjoint problem approach to the identification of an unknown diffusion coefficient,” Inverse Problems, vol. 20, no. 2, pp. 601–625, 2004. View at Publisher · View at Google Scholar · View at MathSciNet
  33. P. DuChateau, “An adjoint method for proving identifiability of coefficients in parabolic equations,” Journal of Inverse and Ill-Posed Problems, vol. 21, no. 5, pp. 639–663, 2013. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  34. A. Hasanov, P. DuChateau, and B. Pektas, “An adjoint problem approach and coarse-fine mesh method for identification of the diffusion coefficient in a linear parabolic equation,” Journal of Inverse and Ill-Posed Problems, vol. 14, no. 5, pp. 435–463, 2006. View at Publisher · View at Google Scholar · View at MathSciNet
  35. Y. H. Ou, A. Hasanov, and Z. H. Liu, “Inverse coefficient problems for nonlinear parabolic differential equations,” Acta Mathematica Sinica, vol. 24, no. 10, pp. 1617–1624, 2008. View at Publisher · View at Google Scholar · View at MathSciNet
  36. L. Gongsheng and M. Yamamoto, “Stability analysis for determining a source term in a 1-D advection-dispersion equation,” Journal of Inverse and Ill-Posed Problems, vol. 14, no. 2, pp. 147–155, 2006. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  37. G. Li, “Data compatibility and conditional stability for an inverse source problem in the heat equation,” Applied Mathematics and Computation, vol. 173, no. 1, pp. 566–581, 2006. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  38. G. Li, D. Yao, and F. Yang, “An inverse problem of identifying source coefficient in solute transportation,” Journal of Inverse and Ill-Posed Problems, vol. 16, no. 1, pp. 51–63, 2008. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  39. G. Li, X. Jia, and C. Sun, “A conditional Lipschitz stability for determining a space-dependent source coefficient in the 2D/3D Advection-dispersion equation,” Journal of Inverse and Ill-Posed Problems, vol. 25, no. 2, pp. 221–236, 2017. View at Publisher · View at Google Scholar · View at MathSciNet
  40. R. C. MacCamy, V. J. Mizel, and T. I. Seidman, “Approximate boundary controllability for the heat equation,” Journal of Mathematical Analysis and Applications, vol. 23, pp. 699–703, 1968. View at Publisher · View at Google Scholar · View at MathSciNet · View at Scopus
  41. R. P. Sperb, Maximum Principles and Their Applications, vol. 157 of Mathematics in Science and Engineering, Academic Press, New York, NY, USA, 1981. View at MathSciNet