Existence of solution to parabolic equations with mixed boundary condition on non-cylindrical domains (Q1753232)

This paper deals with the existence of weak solutions \(y\) to an initial and mixed boundary value problem of a semilinear parabolic equation in non-cylindrical domain \(Q=\bigcup_{t\in (0,T)}\Omega(t)\times\{t\}\subset \mathbb{R}^N\times \mathbb{R}\). The proof is based on the existence of a weak solution \(u\) of an equivalent problem, which is stated under the transformation \(y\mapsto u=\exp[ k_1t+k_2\varphi(x,t)] y\), where \(\varphi\in C^{2,1}(\overline{Q})\) is such that \(\varphi>0\) in \(Q\), and \(\varphi=0\) and \(-\nabla\varphi\cdot \mathbf{n}>\eta>0\) on \(\Sigma=\bigcup_{t\in (0,T)}\partial\Omega(t)\times\{t\}\). The solution \(u\) is established as \(u =w+\bar u\), with \(w\) being the limit of a subsequence of the unique approximated solutions \(w^m\) to the variational problems \(\langle A_mw,v\rangle=\langle L,v\rangle\), \(m\in \mathbb{N}\), and \(\bar u\in H^1(Q)\) taking for the Dirichlet boundary condition on a nonempty set \(\Sigma_0=\mathrm{int}\left(\Sigma \setminus \overline\Sigma_1\right) \). Here, setting \(u=w+\bar u\), \(\langle A_mw,v\rangle=(1/m)\int_Q\partial_t u\partial_t v\mathrm{dxdt}- \int_Q u\partial_t v\mathrm{dxdt}+ \int_Q(a\nabla u)\cdot\nabla v\mathrm{dxdt}+ \int_Q (B\cdot\nabla u) v\mathrm{dxdt}+ \int_QC(\cdot,u) v\mathrm{dxdt}+\int_{\Sigma_1} \left(\cos(\widehat{v,t})+K\right)uv\mathrm{d}\sigma+ (u(\cdot,T), v(\cdot,T))_{\Omega(T)}\) and \(\langle L,v\rangle=(u_0, v(\cdot,0))_{\Omega(0)}+\int_QGv\mathrm{dxdt}+\int_{\Sigma_1} Fv \mathrm{d}\sigma\). For each \(m\in \mathbb{N}\), the existence of \(w^m\) is proved under the standard assumptions, in particular, \(a_{ij}=a_{ji}\in W^1_\infty(Q)\), \(B_i\in L_\infty(Q)\), \(i,j=1,2,\cdots,N\), \(K\in L_\infty(\Sigma_1)\), \(k_2>0\) as \(K\geq 1/2\), \(k_1\) is a negative number small enough independently of \(m\), and either \(C(\cdot,r)\) is Lipschitz continuous with respect to \(r\) and \(\Sigma_0=\Gamma_0\times (0,T)\subset \partial \Omega(0)\times (0,T)\), or \(C(\cdot,r)=C(\cdot)r\).