Optimal control of a phase field system modelling tumor growth with chemotaxis and singular potentials (Q2041028)

The authors consider the coupled parabolic system written as \(\alpha \partial _{t}\mu +\partial _{t}\varphi -\Delta \mu =(P\sigma -A-u)h(\varphi ) \), \(\mu =\beta \partial _{t}\varphi -\Delta \varphi +F^{\prime }(\varphi )-\chi \sigma \), \(\partial _{t}\sigma -\Delta \sigma =-\chi \Delta \varphi +B(\sigma _{s}-\sigma )-D\sigma h(\varphi )+w\), and posed in \(\Omega \times (0,\infty )\), where \(\Omega \subset \mathbb{R}^{3}\) is an open, bounded and connected set with smooth boundary \(\Gamma \). Here \(\varphi \) is an order parameter usually taken between -1 and 1, which represents the healthy cell case and the tumor phase, respectively, \(\mu \) is the chemical potential for \(\varphi \), \(\sigma \) is the nutrient extra-cellular water concentration, which is usually normalized between 0 and 1, \(\alpha \) and \( \beta \) are positive constants, \(A\), \(B\), \(D\), \(P\), \(\chi \) are positive coefficients which represent the apoptosis rate, nutrient supply rate, nutrient consumption rate, proliferation rate, and chemotaxis coefficient, respectively, \(h\) is an interpolation function between -1 and 1 such that \( h(-1)=0\) and \(h(1)=1\), \(F^{\prime }\) is the derivative of a double-well function, and \(u\) and \(w\) are control variables. The homogeneous Neumann boundary conditions \(\partial _{n}\mu =\partial _{n}\varphi =\partial _{n}\sigma =0\) are imposed on \(\partial \Omega \times (0,\infty )\) and \( (\mu ,\varphi ,\sigma )\) start from an initial data \((\mu _{0},\varphi _{0},\sigma _{0})\) at \(t=0\). The authors then consider the energy functional \(\mathcal{J}(\varphi ,\sigma ,u,w)=\frac{\gamma _{1}}{2}\int_{\Omega }\left\vert \varphi (T)-\varphi _{\Omega }\right\vert ^{2}+\frac{\gamma _{2} }{2}\int_{\Omega }\left\vert \varphi (T)-\varphi _{Q}\right\vert ^{2}+\frac{ \gamma _{3}}{2}\int_{\Omega }\left\vert \sigma (T)-\sigma _{\Omega }\right\vert ^{2}+\frac{\gamma _{4}}{2}\int_{\Omega }\left\vert \sigma (T)-\sigma _{Q}\right\vert ^{2}+\frac{\gamma _{5}}{2}\int_{\Omega \times (0,\infty )}u^{2}+\frac{\gamma _{6}}{2}\int_{\Omega \times (0,\infty )}w^{2}\) and the set of admissible controls \(\mathcal{U}_{ad}=\{(u,w)\in (L^{\infty }(\Omega \times (0,\infty )))^{2}:0\leq u_{\ast }\leq u\leq u^{\ast }\), a.e. in \(\Omega \times (0,\infty )\), \(w_{\ast }\leq w\leq w^{\ast }\), a.e. in \( \Omega \times (0,\infty )\}\), where \(u_{\ast },u^{\ast },w_{\ast },w^{\ast }\) are given functions in \(L^{\infty }(\Omega \times (0,\infty ))\). Assuming different hypotheses on the data, the authors first prove the existence of a unique solution \((\mu ,\varphi ,\sigma )\) to the coupled parabolic problem with \(\varphi \in W^{1,\infty }(0,T;L^{2}(\Omega ))\cap H^{1}(0,T;H^{1}(\Omega ))\cap L^{\infty }(0,T;W)\), \(\mu \),\(\sigma \in H^{1}(0,T;L^{2}(\Omega ))\cap L^{\infty }(0,T;H^{1}(\Omega ))\cap L^{2}(0,T;W)\cap L^{\infty }(\Omega \times (0,\infty ))\), where \(W=\{v\in H^{2}(\Omega ):\partial _{n}v=0\) on \(\Gamma \}\). This solution \((\mu ,\varphi ,\sigma )\) satisfies further properties and the authors prove that this solution continuously depends on the control variables \((u,w)\). Finally, the existence of a least one solution to the control problem is proved and the authors derive first-order necessary optimality conditions. For the existence and uniqueness result, the authors apply a Faedo-Galerkin method and they use a Yosida regularization of the derivative function \( F^{\prime }\). For the existence of a solution to the control problem, the authors mainly use weak compactness arguments. The authors also prove that the linearized system deduced from the above parabolic system has a unique solution \((\eta ,\xi ,\zeta )\) which belongs to \(H^{1}(0,T;L^{2}(\Omega ))\cap L^{\infty }(0,T;H^{1}(\Omega ))\cap L^{2}(0,T;W)\) and that the control-to-state operator \(\mathcal{S}\) is Fréchet differentiable as a mapping from \((L^{2}(\Omega \times (0,\infty )))^{2}\) into an appropriate Banach space \(\mathcal{Y}\).