1 Introduction

Let \(\Omega = ]0, + \infty [ \times \Omega _0\), where \(\Omega _0\) is an open set of \(\mathbb {R}^2\) with enough smooth boundary (for example \(\Omega _0\) is of class \(C^m\), with \(m \ge 2\)). Let us set \(S= [0, + \infty [ \times \partial \Omega _0\), where \( \partial \Omega _0\) is the boundary of \(\Omega _0\). Let us consider the following class of hyperbolic second-order operators with double characteristics in presence of transition:

$$\begin{aligned} P=D^2_{x_0} - D^2_{x_1} - (x_0- \alpha (x_1,x_2))^2 D^2_{x_2} + \sum _{j=0}^2 a_j(x) D_{x_j} +b(x), \quad \mathrm{in} \ \Omega , \end{aligned}$$
(1)

where \(x = (x_0, x_1, x_2 )\), \(\mathrm{Im} \, a_2(x)=(x_0-\alpha (x'))\widetilde{a}_2(x)\), with \(\widetilde{a}_2(x)\) real function, \(D_{x_j}= \dfrac{1}{i} \partial _{x_j}\), \(j=0,1,2\), the coefficients belong in \(C^{\infty }(\widetilde{\Omega })\), \(\widetilde{\Omega } =[0, + \infty [ \times \widetilde{\Omega }_0\), with \(\widetilde{\Omega }_0\) an open set containing strictly \(\Omega _0\), and \(\alpha \) is a real function. Let \(x'=(x_1,x_2)\), \(\xi =(\xi _0, \xi _1, \xi _2)=(\xi _0, \xi ')\), where we set \(\xi '= (\xi _1, \xi _2)\). Let

$$\begin{aligned} p(x_0, x', \xi ) = - \xi _0^2+ \xi ^2_1+ (x_0 - \alpha (x'))^2 \xi ^2_2 + \frac{1}{i} \sum _{j=0}^2 a_{j}(x) \xi _j + b(x) \end{aligned}$$

be the symbol of P, let

$$\begin{aligned} \Sigma = \left\{ \rho = (x_0, x', \xi ) \in T^* \Omega : \ p(\rho ) =0, \ \nabla p(\rho ) =0 \right\} \end{aligned}$$

be the characteristic set and let

$$\begin{aligned} F_p(\rho )=\frac{1}{2} \left( \begin{array}{cc} p''_{x\xi }(\rho ) &{} p''_{\xi \xi }(\rho ) \\ - p''_{x x}(\rho ) &{} - p''_{\xi x}(\rho ) \end{array} \right) , \quad \forall \rho \in \Sigma \end{aligned}$$

be the fundamental matrix of P at \(\rho \). The spectrum of \(F_p(\rho )\), denoted by \(\mathrm{Spec}(F_p(\rho ))\), has an important rule to study the well-posedness of the Cauchy–Dirichlet problem associated to the operator P. In particular, it results (see [10])

$$\begin{aligned} z \in \mathrm{Spec}(F_p(\rho )) \ \Leftrightarrow \ - z, \overline{z} \in \mathrm{Spec}(F_p(\rho )). \end{aligned}$$

The fundamental matrix of P at \(\rho \) has only pure imaginary eigenvalues with a possible exception of a pair of nonzero real eigenvalues \(\pm \lambda \) (see [9,10,11]). If \(F_p(\rho )\) has a pair of nonzero real eigenvalues, P is called effectively hyperbolic at \(\rho \). If \(F_p(\rho )\) has only pure imaginary eigenvalues and if there are only Jordan blocks of dimension 2 in the Jordan normal form of \(F_p(\rho )\) corresponding to the eigenvalue 0, i.e., \(\mathrm{Ker} F_p(\rho )^2 \cap \mathrm{Im} F_p(\rho )^2 = \{ 0\}\), P is called non-effectively hyperbolic of type 1 at \(\rho \). Instead, if \(F_p(\rho )\) has only pure imaginary eigenvalues and if there is only a Jordan block of dimension 4 and no block of dimension 3 in the Jordan normal form of \(F_p(\rho )\) corresponding to the eigenvalue 0, i.e., \(\mathrm{Ker} F_p(\rho )^2 \cap \mathrm{Im} F_p(\rho )^2\) is 2-dimensional, P is called non-effectively hyperbolic of type 2 at \(\rho \). Furthermore, let

$$\begin{aligned} \Sigma _+= & {} \left\{ \rho \in \Sigma : \ P \ \mathrm{is \ effectively \ hyperbolic \ at} \ \rho \right\} , \\ \Sigma _-= & {} \left\{ \rho \in \Sigma : \ P \ \mathrm{is \ non-effectively \ hyperbolic \ of \ type} \ 1 \ \mathrm{at} \ \rho \right\} , \\ \Sigma _0= & {} \left\{ \rho \in \Sigma : \ P \ \mathrm{is \ non-effectively \ hyperbolic \ of \ type} \ 2 \ \mathrm{at} \ \rho \right\} , \end{aligned}$$

(see [9]). It is easy to deduce

$$\begin{aligned} \Sigma = \Sigma _- \sqcup \Sigma _0 \sqcup \Sigma _+. \end{aligned}$$

We say that we have a transition exactly when at least two among the above sets are nonempty.

The paper continues the study on the following Cauchy–Dirichlet problem

$$\begin{aligned} {\left\{ \begin{array}{ll} Pu=f, \quad \mathrm{in} \ \Omega =]0,+ \infty [ \times \Omega _0 \\ u|_{\partial \Omega } =0, \ \frac{du}{d n}|_{\Omega _0} =0, \ u|_S=0 \end{array}\right. } \end{aligned}$$
(2)

started in [7]. In fact, in [7], several a priori estimates of local or global nature in Sobolev spaces with general exponent \(s \le 0\) for the class of second-order hyperbolic operators (1) are proved. Here, we establish some existence results for the Cauchy–Dirichlet problem (2). To this aim, we need to obtain other a priori estimates in Sobolev spaces with exponent \(s \le 0\). The proofs of such estimates make use of delicate variational techniques because of the degeneration on the characteristic set and of the transition between \(\Sigma _-\), \(\Sigma _0\) and \(\Sigma _+\). More precisely, the function \(\alpha \) in (1) depends on the variables \(x_1\) and \(x_2\). As a consequence, the coefficient \(x_0 - \alpha (x')\) degenerates on the characteristic set with respect to all the variables. Setting \(\beta = x_0 -\alpha (x')\), if \(| \partial _{x_1} \alpha (x')|<1\), \(\beta =0\) and \(\xi _0= \xi _1=0\), then \(F_p(\rho )\) has two distinct nonzero real eigenvalues. If \(| \partial _{x_1} \alpha (x')| > 1\), \(\beta =0\) and \(\xi _0= \xi _1=0\), \(F_p(\rho )\) has two nonzero imaginary eigenvalues. In conclusion, let \(\overline{\Sigma }\) be the set of points \(\rho =(x_0,x',\xi )\) of \(\Sigma \) such that \(\beta =0\) and \(\xi _0=\xi _1=0\). We have that \(\rho \in \Sigma _+\) if \(\rho \in \overline{\Sigma }\) and \(| \partial _{x_1} \alpha (x')|<1\), \(\rho \in \Sigma _-\) if \(\rho \in \overline{\Sigma }\) and \(| \partial _{x_1} \alpha (x')|>1\), and \(\rho \in \Sigma _0\) if \(\rho \in \overline{\Sigma }\) and \(| \partial _{x_1} \alpha (x')|=1\). Hence, even if we study the special class of operators (1), the transition from effectively hyperbolic to non-effectively hyperbolic occurs. A class more general of hyperbolic second-order operators with double characteristics is analyzed in [6]. It is worth to underline that the coefficient \(x_0 - \alpha (x')\) does not contain the parameter \(\lambda \) very helpful to prove global estimates near the boundary of \(\Omega \) in [5]. Finally, we remark that the operator (1) contains the first-order terms and the zero-order term, which have an important rule to study the well-posedness of the problem. Instead in [4], the subprincipal term is identically zero; consequently, the Hörmander–Ivrii–Petkov condition is automatically verified.

Several scholars considered the Cauchy problem either for effectively or non-effectively hyperbolic operators with double characteristics (see, for instance, [8, 10,11,12,13,14,15,16]). In [9], another class of hyperbolic second-order operators with double characteristics is analyzed. In particular, the \(C^{\infty }\) well-posedness of the Cauchy problem and Carleman estimates for non-effectively hyperbolic operators have been obtained. In [17], some energy estimates for a different class of hyperbolic second-order operators are established. Moreover, the \(C^{\infty }\) well-posedness of the Cauchy problem for non-effectively hyperbolic operators is studied. We underline that in [9, 17] the Cauchy problem for a class of operators in a form more general then (1) is analyzed, but a priori estimates only when \(\Sigma = \Sigma _- \sqcup \Sigma _0\) are established. Instead, thanks to variational and pseudodifferential techniques different from the ones used in [9, 17], we are able to examine the mixed Cauchy–Dirichlet problem and we prove a priori estimates when \(\Sigma = \Sigma _- \sqcup \Sigma _0 \sqcup \Sigma _+\) or \(\Sigma = \Sigma _- \sqcup \Sigma _0\) or \(\Sigma = \Sigma _0 \sqcup \Sigma _+\) or \(\Sigma = \Sigma _-\) or \(\Sigma = \Sigma _+\). Moreover, in the class of operators (1), studied also in [1,2,3], both the case in which \(F_p(\rho )\) has two distinct real eigenvalues and the case in which all the eigenvalues are purely imaginary numbers can occur.

We set \(\beta (x)=x_0-\alpha (x')\), \(g(x')= \frac{\alpha (x')}{\partial _{x_1} \alpha (x')}\), \(h(x')=1-\partial _{x_1} g(x')\), in \(\widetilde{\Omega }\),

$$\begin{aligned} \Gamma= & {} \{ x \in \widetilde{\Omega }: \ \beta (x)=0 \},\\ \Gamma '= & {} \{ x \in \Gamma : \ \alpha (x') \ge 0 \},\\ \Omega _0'= & {} \{ x' \in \widetilde{\Omega }_0: \ \alpha (x') \ge 0 \}. \end{aligned}$$

Moreover, let \(B= (b_{hk})_{h,k=0,1}\) be the quadratic matrix-function whose elements are given by:

$$\begin{aligned} b_{00}(x) &= h(x') -2 \alpha (x') \widetilde{a}_0(x), \quad \forall x \in \widetilde{\Omega }, \\ b_{01}(x)& = b_{10}(x) = - g(x') \widetilde{a}_0(x) - \alpha (x') \widetilde{a}_1(x), \quad \forall x \in \widetilde{\Omega }, \\ b_{11}(x)& = h(x') -2 g(x') \widetilde{a}_1(x), \quad \forall x \in \widetilde{\Omega }, \end{aligned}$$

where \(\widetilde{a}_0\) and \(\widetilde{a}_1\) are the imaginary parts of \(a_0\) and \(a_1\), respectively.

We suppose

  1. (i)

    \(g,h \in C^{\infty }(\Omega '_0)\), \(h(x') \in [h_1, h_2]\), \(\forall x' \in \Omega '_0\), with \(0<h_1<h_2 < 4\);

  2. (ii)

    the matrix-function B is positive definite in \(\Gamma '\), namely there exists \(k>0\) such that \(B(x') \eta \cdot \eta \ge k \Vert \eta \Vert ^2\), \(\forall \eta = (\eta _1, \eta _2) \ne (0,0)\), \(\forall x \in \Gamma '\);

  3. (iii)

    \(g(x') n_1|_{S} \ge 0\), for every \(x' \in \Omega _0' \cap \partial \Omega _0\).

We remark that if \(\widetilde{a}_0 = \widetilde{a}_1 =0\), on \(\Gamma '\), assumption (ii) is verified.

The main goal of the paper is to prove the following results:

Theorem 1

Let (i), (ii) and (iii) be satisfied. If \(f \in L^{2}_{loc}(\overline{\Omega })\), there exists \(w \in L^2_{loc}(\overline{\Omega })\) such that

$$\begin{aligned} (w, \, ^t Pu) = (f,u), \quad \forall u \in C_0^{\infty }(\overline{\Omega }): \ u|_S=0, \end{aligned}$$

where \(\overline{\Omega } = [0, + \infty [ \times \overline{\Omega }_0\).

Theorem 2

Let (i), (ii) and (iii) be satisfied. Let \(f \in H^{r}_{loc}(\overline{\Omega })\), with \(r \ge 2\), the Cauchy–Dirichlet problem

$$\begin{aligned} {\left\{ \begin{array}{ll} Pu=f, \quad \mathrm{in} \ \Omega =]0,+ \infty [ \times \Omega _0 \\ u|_{\partial \Omega } =0, \ \frac{du}{d n}|_{\Omega _0} =0, \ u|_S=0 \end{array}\right. } \end{aligned}$$

admits a solution \(u \in H^{r}_{loc}(\overline{\Omega } \setminus \partial \Omega _0)\).

Let us consider some operators which satisfy assumptions (i), (ii) and (iii) and for which we have a transition.

Example 1

Let \(\alpha (x')= x_1^3 e^{kx_2}\) be functions in an open set \(\widetilde{\Omega }_0\) of \(\mathbb {R}^2\) contained (0, 0). Let \(P= D_{x_0}^{(2)} - D_{x_1}^{(2)} - (x_0-\alpha (x'))^2 D_{x_2}^{(2)} - i a_0 D_{x_0}\), where \(a_0 >0\). It results \(g(x') = \frac{1}{3}x_1\) and \(h(x') = \frac{2}{3}\), then assumption (i) is verified for every \(\widetilde{\Omega }_0\). Assumption (ii) is satisfied for every \(\widetilde{\Omega }_0 \subseteq ]-\infty , \frac{2}{a_0} ] \times \mathbb {R}\). Moreover, assumption (iii) is fulfilled if \(n_1\) on \(\partial \Omega _0 \cap \Omega _0'\) is positive (for example if \(\widetilde{\Omega }_0\) is a circle of center in (0, 0)). Then, we can choose \(\widetilde{\Omega }_0\) such that \(|\partial _{x_1} \alpha (x')|\) admits values either less than or equal than or greater than 1. As a consequence, it follows \(\Sigma = \Sigma _- \sqcup \Sigma _0 \sqcup \Sigma _+\), with \(\Sigma _-\), \(\Sigma _0\) and \(\Sigma _+\) nonempty.

Example 2

Now, let us consider \(\alpha (x')= (ax_1+bx_2+c)^2\), with \(a,b,c \in \mathbb {R}\), \(a,b \ne 0\), in an open set \(\widetilde{\Omega }_0\) of \(\mathbb {R}^2\) contained (0, 0). Let \(P= D_{x_0}^{(2)} - D_{x_1}^{(2)} - (x_0-\alpha (x'))^2 D_{x_2}^{(2)} +a_0 D_{x_0} - i a_1(x) (x_0-\alpha (x')) (D_{x_1} + D_{x_2})\), where \(a_0 \in \mathbb {R}\) and \(a_1 \in C^{\infty }\). It results \(g(x')= \dfrac{ax_1+bx_2+c}{2 a}\) and \(h(x')= \dfrac{1}{2}\). Hence, assumption (i) is always verified. Moreover, we can choose \(\widetilde{\Omega }_0\) such that assumption (iii) is fulfilled and both \(|\partial _{x_1} \alpha (x')| \le 1\) and \(|\partial _{x_1} \alpha (x')| \ge 1\) hold. Therefore, the existence of a solution is ensured in presence of transition.

The paper is organized as follows. In Sect. 2, some preliminary notations are recalled. In Sect. 3, a priori estimates obtained in [7] are referred. Section 4 is devoted to prove a priori estimates under the assumption \(|\partial _{x_1} \alpha (x')| \le 1\). Instead, Sect. 5 concerns estimates under the assumption \(|\partial _{x_1} \alpha (x')| \ge 1\). In Sect. 6, conclusive estimates in \(L^2\) are proved. In Sect. 7, estimates in Sobolev spaces with \(s<0\) are established making use of the pseudodifferential operator theory. Section 8 concerns the study of some global estimates. Finally, Sects. 9 and 10 deal with the proofs of Theorems 1 and 2, respectively.

2 Notations and preliminaries

Let \(\alpha =(\alpha _0, \alpha _1, \alpha _2) \in \mathbb {N}^3_0\). We indicate the derivative of order \(|\alpha |\) by \(\partial ^{\alpha }\), the derivative of order h with respect to \(x_j\) by \(\partial ^{h}_{x_j}\) and the derivative of order h with respect to \(x_j\) and \(x_p\) by \(\partial ^{h}_{x_j, x_p}\).

We indicate the \(L^2\)-scalar product, the \(L^2\)-norm and the \(H^{r}\)-norm by \((\cdot , \cdot )\), \(\Vert \cdot \Vert \) and \(\Vert \cdot \Vert _{H^{r}}\) (\(r \in \mathbb {N}_0\)), respectively. We indicate the external normal versor to the boundary \(\partial \Omega \) by \(n=(n_0,n_1,n_2)\).

Let \(C_0^{\infty }(\overline{\Omega })\) be the space of restrictions of functions belonging to \(C^{\infty }_0(\mathbb {R}^3)\) on \(\overline \Omega \). For each \(K \subseteq \overline{\Omega }\) compact set, let \(C^{\infty }_0 (K)\) be the set of functions \(\varphi \in C^{\infty }_0 (\overline{\Omega })\) having support contained in K. Set \(\Omega _k=[0,k[ \times \Omega _0\), let us introduce

$$\begin{aligned} C^{\infty }_0(\overline{\Omega }_k) = \left\{ u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq [0,k[ \times \overline{\Omega }_0 \right\} . \end{aligned}$$

Moreover, let \(C^{*\infty }_0(\overline{\Omega })\) be the space of functions \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\gamma _1 \partial _{x_0} u(0,x') = \gamma _2 u(0,x')\), where \(\gamma _1, \gamma _2 \in \mathbb {R}\). Consequently, we can introduce \(C^{* \infty }(\widetilde{\Omega })\) and \(C^{* \infty }(\overline{\Omega }_k)\). It is worth to remark that if \(u \in C^{\infty }_0(\Omega )\), then \(u \in C^{*\infty }_0(\overline{\Omega }_0)\). Furthermore, if \(u(x_0,x')=u_1(x') u_2(x_0)\), with \(u_1 \in C^{\infty }_0(\overline{\Omega }_0)\), \(u|_{\partial \Omega _0}=0\) and \(u_2 \in C^{\infty }_0([0,k[)\) then \(u \in C^{*\infty }_0(\overline{\Omega }_k)\).

Let \(S(\mathbb {R}^3)\) be the space of rapidly decreasing functions. Let \(S(\overline{\Omega })\) be the space of restrictions of functions belonging to \(S(\mathbb {R}^3)\) on \(\overline{\Omega }\).

Let \(\Omega = ]0, + \infty [ \times \Omega _0\) and let \(s \in \mathbb {R}\), the norm in \(H^{0,s}\) is given by

$$\begin{aligned} \Vert u \Vert ^2_{H^{0,s}}= & {} \frac{1}{(2 \pi )^2} \int _0^{+ \infty } {\text {d}}x_0 \int _{\mathbb {R}^2} (1+|\xi '|^2)^{s} | \widehat{u}(x_0, \xi ')|^2 {\text {d}}\xi ', \\&\quad \forall u \in C^{\infty }_{0}(\overline{\Omega }): \ \mathrm{supp} \ u \subseteq [0, + \infty [ \times \Omega _0, \end{aligned}$$

where the Fourier transform is done only with respect to the variable \(x'\). Let \(A_s: C^{\infty }_0(\Omega ) \rightarrow C^{\infty }(\Omega )\) be the pseudodifferential operator defined by

$$\begin{aligned} A_s u= & {} \frac{1}{(2 \pi )^2} \int _{\mathbb {R}^2} e^{i x' \cdot \xi '} (1+|\xi '|^2)^{\frac{s}{2}} \widehat{u}(x_0, \xi ') {\text {d}}\xi ', \\&\quad \forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \ u \subseteq [0, + \infty [ \times \Omega _0. \end{aligned}$$

For every \(\varphi (x') \in C^{\infty }_0(\Omega _0)\), the operator \(\varphi A_s u\) extends as a linear continuous operator from \(H^{0,r}_{comp.}(\Omega )\) into \(H^{0,r-s}_{loc}(\Omega )\), where \(r,s \in \mathbb {R}\). In particular, in \(\Omega _k =[0,k[ \times \Omega _0\), for \(k>0\), let \(H^{0,s}(\Omega _k)\) be the space of \(u \in H^{0,s}(\Omega _k)\) such that \(\mathrm{supp} \; u \subseteq \Omega _k\). Moreover, if \(\mathrm{supp} \ \varphi \subseteq \Omega _0 \setminus \mathrm{supp} \ u\), then \(\varphi A_s u\) is a regularizing operator with respect to the variable \(x'\). It results

$$\begin{aligned} \Vert \varphi A_s u \Vert _{H^{0,r}} \le c \Vert u \Vert _{H^{0,r'}}, \quad \forall r,r' \in \mathbb {R}, \ u \in C^{\infty }(\overline{\Omega }): \ \mathrm{supp} \ u \subseteq [0, + \infty [ \times \Omega _0. \end{aligned}$$

The norms \(\Vert u \Vert _{H^{0,s}(\Omega )}\) and \(\Vert A_s u \Vert _{L^2(\Omega )}\) are equivalent for any \(s \in \mathbb {R}\).

Let \(s \in \mathbb {R}\) and \(p \ge 0\). Let \(H^{p,s}(\mathbb {R}^3)\) be the space of all the distributions on \(\mathbb {R}^3\) such that

$$\begin{aligned} \Vert u \Vert ^2_{H^{p,s}(\mathbb {R}^3)} = \frac{1}{(2 \pi )^2} \sum _{|h| \le p} \int _{\mathbb {R}^3} (1+|\xi '|^2)^{s} | \partial ^h_{x_0} \widehat{u}(x_0, \xi ')|^2 {\text {d}}x_0 {\text {d}}\xi ' < + \infty . \end{aligned}$$

Let \(H^{p,s}(\Omega )\) be the space of restrictions of elements of \(H^{p,s}(\mathbb {R}^3)\) on \(\Omega \) endowed with the norm

$$\begin{aligned} \Vert u \Vert _{H^{p,s}(\Omega )} = \inf _{{\begin{array}{c} U \in H^{p,s}(\mathbb {R}^3) \\ U|_{\Omega } =u \end{array}}} \Vert U \Vert _{H^{p,s}(\mathbb {R}^3)}. \end{aligned}$$

In the same way, the space \(H^{p,s}(\Omega _k)\) can be introduced.

At last, we consider the transposed operator of the operator P:

$$\begin{aligned} ^tP= & {} - \partial _{x_0}^2 + \partial _{x_1}^2 + (x_0-\alpha (x'))^2 \partial ^2_{x_2} - 4 (x_0-\alpha (x')) (\partial _{x_2} \alpha ) \partial _{x_2} \\&\quad - \frac{1}{i} \sum _{j=0}^2 a_j(x) \partial _{x_j} - \frac{1}{i} \sum _{j=0}^2 \partial _{x_j} a_j(x) - 2 (\partial _{x_2} \alpha )^2 +b(x). \end{aligned}$$

3 Some known preliminary results

First of all, we recall a priori estimate for the solution to the problem (2) (see [2], Lemma 3.1).

Lemma 1

Let \(u \in S(\overline{\Omega })\) and let \(p, \alpha _0, \alpha _1, \alpha _2 \in \mathbb {N}_0\). Then

$$\begin{aligned} \Vert x_0^{\frac{p}{2}} \partial ^{\alpha _0, \alpha _1, \alpha _2} u \Vert \le \frac{2}{p+1} \Vert x_0^{\frac{p+2}{2}} \partial ^{\alpha _0+1, \alpha _1, \alpha _2} u \Vert . \end{aligned}$$
(3)

Moreover, we have the following preliminary result (see [7], Lemma 3.2).

Lemma 2

Let \(u \in S(\overline{\Omega })\), it results

$$\begin{aligned} \int _{\Omega _0} | u(0,x')|^2 {\text {d}}x' \le 4 \Vert x_0 \partial _{x_0} u \Vert \Vert \partial _{x_0}u \Vert . \end{aligned}$$

The next result holds (see [7], Lemma 3.3).

Lemma 3

For every \(\varepsilon , \delta >0\) there exists \(k>0\) such that, if

$$\begin{aligned} I_{k,\delta } = \left\{ x \in \overline{\Omega }: \ x_0 < k, \ |x_0 - \alpha (x') | > \delta \right\} , \end{aligned}$$

it results

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \ ^{t} Pu \Vert ,\nonumber \\&\qquad \forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \ u \subseteq I_{k, \delta }, \ u|_S=0. \end{aligned}$$
(4)

We present a priori estimate (see [7], Theorem 3.4).

Theorem 3

Let (i) and (iii) be satisfied. Then, there exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \ ^{t} Pu\Vert , \nonumber \\&\qquad \forall u \in C^{*\infty }_0(\overline{\Omega }_k): \ u|_S=0. \end{aligned}$$
(5)

Moreover, we recall the following result (see [7], Theorem 3.5).

Theorem 4

Let (i) and (iii) be satisfied. For every \(\varepsilon >0\) there exist \(k>0\) and a neighborhood \(I_{x'}\) in \(\Omega _0 \cap \Gamma \) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 -\alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^tP u \Vert , \\&\qquad \forall u \in C^{*\infty }_0 (\Omega _k): \ \mathrm{supp} \, u \subseteq [0,k[ \times I_{x'}, \ u|_S=0. \end{aligned}$$

Let \(\overline{x}_0 >0\) and let \(k>0\), we denote by \(\Omega _{\overline{x}_0,k}=] \overline{x}_0, \overline{x}_0+ k [ \times \overline{\Omega }_0\). Let us show the following preliminary result (see [7], Lemma 4.1).

Lemma 4

Let \(u \in S(\Omega )\) such that \(\partial _{x_0} u|_{\Omega _0} =0\), let \(p, \alpha _0, \alpha _1, \alpha _2 \in \mathbb {N}\) and \(\overline{x}_0 >0\). It results

$$\begin{aligned} \Vert (x_0 - \overline{x}_0)^{\frac{p}{2}} \partial ^{\alpha _0, \alpha _1, \alpha _2} u \Vert \le \frac{2}{p+1} \Vert (x_0 - \overline{x}_0)^{\frac{p+2}{2}} \partial ^{\alpha _0+1, \alpha _1, \alpha _2} u \Vert . \end{aligned}$$

We consider another preliminary lemma (see [7], Lemma 4.2).

Lemma 5

For every \(\varepsilon , \delta > 0\) and \(\overline{x}_0 >0\), there exists \(k>0\) such that, setting

$$\begin{aligned} I_{k, \delta } = \left\{ x \in \overline{\Omega }: \ x_0 \in ] \overline{x}_0, \overline{x}_0+ k [, \ |x_0 - \alpha (x')|> \delta \right\} , \end{aligned}$$

it results

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^t P u \Vert , \\&\qquad \forall u \in C_0^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq I_{k, \delta }, \ u|_{S}=0. \end{aligned}$$

We recall the following preliminary result (see [7], Lemma 4.3).

Lemma 6

Let (i), (ii) and (iii) be satisfied. Let \(\overline{x}_0 >0\), for every \(\varepsilon >0\) there exists \(k, \delta >0\) such that, setting

$$\begin{aligned} J_{k, \delta } = \left\{ x \in \overline{\Omega }: \ x_0 \in ] \overline{x}_0, \overline{x}_0+ k [, \ |x_0 - \alpha (x')|< \delta \right\} , \end{aligned}$$

it results

$$\begin{aligned}&\Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^tPu \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) , \\&\qquad \forall u \in C_0^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq J_{k, \delta }, \ u|_S=0. \end{aligned}$$

At last, we present the following result (see [7], Theorem 4.4).

Theorem 5

Let (i), (ii) and (iii) be satisfied. Let \(\overline{x}_0>0\). There exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 -\alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^tP u \Vert , \\&\qquad \forall u \in C^{\infty }_0 (\Omega _{\overline{x}_0,k}): \ u|_S=0. \end{aligned}$$

4 Estimates under the assumption \(|\partial _{x_1} \alpha (x') | \le 1\)

Let \(\overline{x}_0 \ge 0\), let us denote by

$$\begin{aligned} J_{k, \delta , \overline{x}_0} = \left\{ x \in \overline{\Omega }: \ x_0 \in [ \overline{x}_0, \overline{x}_0 + k[, \ |x_0 - \alpha (x') | < \delta \right\} . \end{aligned}$$

The following result holds.

Theorem 6

Let (i), (ii) and (iii) be satisfied. Let us assume that there exist two positive numbers \(k'\) and \(\delta \) such that \(|\partial _{x_1}\alpha (x')| \le 1\) on \(\Omega _0 \cap J_{k', \delta , 0}\). Then, for every \(\varepsilon >0\) there exists \(0 < k \le k'\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^t Pu \Vert ,\nonumber \\&\quad \, \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{k} = [ 0, k[ \times \overline{\Omega }_0, \ u|_S=0. \end{aligned}$$
(6)

Proof

Let us consider the following inner products

$$\begin{aligned} (\, ^t Pu, x_0 \partial _{x_0}u) + (x_0 \partial _{x_0}u, \, ^t Pu). \end{aligned}$$

By means of integrations by parts, for every \(u \in C_{0}^{\infty }(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq \Omega _{k'} \cap J_{k', \delta , 0}\) and \(u|_S=0\), we have

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \\&\qquad + 2 (x_0 (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_2} u) - 4 (x_0 (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_0} u) \\&\quad = ( \, ^t Pu, x_0 \partial _{x_0} u) + ( x_0 \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, x_0 \partial _{x_0} u) - ( x_0 \partial _{x_0} u, \, ^t (P-P_2)u). \end{aligned}$$

From which it follows

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \frac{1}{2} \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \nonumber \\&\qquad + \left( (x_0-\alpha (x')) \left( \frac{5}{2} x_0 - \frac{1}{2} \alpha (x') \right) \partial _{x_2} u, \partial _{x_2} u \right) \nonumber \\&\qquad - 4 (x_0 (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_0} u) \nonumber \\&\quad = ( \, ^t Pu, x_0 \partial _{x_0} u) + ( x_0 \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, x_0 \partial _{x_0} u) \nonumber \\&\qquad - ( x_0 \partial _{x_0} u, \, ^t (P-P_2)u). \end{aligned}$$
(7)

We denote by

$$\begin{aligned} \Omega _{k', \frac{1}{5}} = \left\{ x \in \Omega _{k'}: \ \frac{1}{5} \alpha (x') \le x_0 \le \alpha (x') \right\} . \end{aligned}$$

Since \((x_0-\alpha (x')) \left( \frac{5}{2} x_0 - \frac{1}{2} \alpha (x') \right) >0\), in \(\Omega _{k'} \setminus \Omega _{k', \frac{1}{5}}\), by (7) one has

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \frac{1}{2} \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \nonumber \\&\quad \le - \left( (x_0-\alpha (x')) \left( \frac{5}{2} x_0 - \frac{1}{2} \alpha (x') \right) \partial _{x_2} u, \partial _{x_2} u \right) _{\Omega _{k^{'}, \frac{1}{5}}} \nonumber \\&\qquad + 4 |(x_0 (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_0} u)| \nonumber \\&\qquad + 2 \Vert x_0 \, ^t Pu \Vert \Vert \partial _{x_0} u \Vert + 2 \Vert x_0 \, ^t (P- P_2) u \Vert \Vert \partial _{x_0} u \Vert , \end{aligned}$$
(8)

where we denoted by \((\cdot , \cdot )_{\Omega _{k^{'}, \frac{1}{5}}}\) the inner product on \(\Omega _{k^{'}, \frac{1}{5}}\). Furthermore, it results

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \\&\quad \le - 2 \left( x_0 (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_2} u \right) _{\Omega _{k^{'}, \frac{1}{5}}} + 4 \Vert x_0 (x_0-\alpha (x')) \partial _{x_2} u \Vert \Vert \partial _{x_0} u \Vert \\&\qquad + 2 \Vert x_0 \, ^t Pu \Vert \Vert \partial _{x_0} u \Vert + 2 \Vert x_0 \, ^t (P- P_2) u \Vert \Vert \partial _{x_0} u \Vert \\&\quad \le 2 \left( \alpha (x') (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_2} u \right) _{\Omega _{k^{'}, \frac{1}{5}}} + 4 \Vert x_0 (x_0-\alpha (x')) \partial _{x_2} u \Vert \Vert \partial _{x_0} u \Vert \\&\qquad + 2 \Vert x_0 \, ^t Pu \Vert \Vert \partial _{x_0} u \Vert + 2 \Vert x_0 \, ^t (P- P_2) u \Vert \Vert \partial _{x_0} u \Vert . \end{aligned}$$

In \(\Omega _{k', \frac{1}{5}}\), we consider the following inner products

$$\begin{aligned} (\partial _{x_0} u, \, ^t Pu) + (\, ^t Pu, \partial _{x_0} u). \end{aligned}$$

If u is identically zero on \(\Gamma _{\eta }\), where \(\Gamma _{\eta }\) is the surface \(x_0 = \eta \alpha (x')\), with \(0 < \eta \le \frac{1}{5}\), integrating by parts, we have

$$\begin{aligned}&2 \Vert (\alpha (x') - x_0)^{\frac{1}{2}} \partial _{x_2} u \Vert ^2 - 4 ( (x_0-\alpha (x')) \alpha (x') \partial _{x_2} u, \partial _{x_0} u) \nonumber \\&\qquad + \int _{\Gamma } \left[ (\partial _{x_0} u)^2 + 2 \partial _{x_1} \alpha (x') \partial _{x_0} u \, \partial _{x_1} u + ( \partial _{x_1} u )^2 \right] {\text {d}} \sigma \nonumber \\&\quad = ( \, ^t Pu, \partial _{x_0} u) + ( \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, \partial _{x_0} u) - ( \partial _{x_0} u, \, ^t (P-P_2)u). \end{aligned}$$
(9)

By (8) and (9), if \(| \partial _{x_1} \alpha (x')| \le 1\), on \(\Omega _0 \cap J_{k', \delta , 0}\), and \(k'\) is small enough, the claim follows assuming that u is identically zero on \(\Gamma _{\eta }\).

Let \(u \in C_{0}^{\infty }(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq \Omega _{k'}\) and \(u|_S=0\). Let \(\chi \) be a function of class \(C^{\infty }\) such that \(\chi (t)=1\), for \(t \ge \eta \), and \(\chi (t)=0\), for \(0 \le t \le \frac{\eta }{2}\). Rewriting (9) for \(u \chi \left( \frac{x_0}{\alpha (x')} \right) \) and adding (8), there exists \(0 < k \le k'\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^t Pu \Vert , \\&\qquad \ \, \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{k} \cap J_{k, \delta , 0}, \ u|_S=0. \nonumber \end{aligned}$$

Making use of the previous inequality and Lemma 3 with k small enough, the claim is achieved. \(\square \)

We set

$$\begin{aligned} \Omega _{\overline{x}_0, k} = [\overline{x}_0, \overline{x}_0+k[ \times \overline{\Omega }_0, \end{aligned}$$

with \(\overline{x}_0 >0\) and \(k>0\), and we prove the following result.

Theorem 7

Let (i), (ii) and (iii) be satisfied. Let us assume that there exist two positive numbers \(k'\) an \(\delta \) such that \(|\partial _{x_1}\alpha (x')| \le 1\), on \(\Omega _{\overline{x}_0} \cap J_{k', \delta , \overline{x}_0}\), where \(\Omega _{\overline{x}_0}\) is the part of the plane \(x_0= \overline{x}_0\) in \(\Omega _{\overline{x}_0,k}\). Then, for every \(\varepsilon >0\) there exists \(0< k \le k'\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^t Pu \Vert ,\nonumber \\&\qquad \ \ \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0, k}, \ u|_S=0. \end{aligned}$$
(10)

Proof

Let \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0, k'} \cap J_{k', \delta , \overline{x}_0}\) and \(u|_S=0\), integrating by parts in the following inner products

$$\begin{aligned} (\, ^t Pu, (x_0 - \overline{x}_0) \partial _{x_0} u) + ( (x_0 - \overline{x}_0) \partial _{x_0} u, \, ^t Pu), \end{aligned}$$

we obtain

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \\&\qquad + 2 ((x_0 - \overline{x}_0) (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_2} u) + 4 (x_0 (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0} u) \\&\quad = ( \, ^t Pu, (x_0 - \overline{x}_0) \partial _{x_0} u) + ( (x_0 - \overline{x}_0) \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, (x_0 - \overline{x}_0) \partial _{x_0} u) \\&\qquad - ( (x_0 - \overline{x}_0) \partial _{x_0} u, \, ^t (P-P_2)u), \quad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k}, \ u|_S=0. \end{aligned}$$

Taking into account that \(\frac{1}{2} \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 + 2 ((x_0 - \overline{x}_0) (x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_2} u) < 0\), if \(x_0 \le \frac{1}{5} \alpha (x') + \frac{4}{5} \overline{x}_0\) or \(x_0 \ge \alpha (x')\), it results

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \frac{1}{2} \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 \nonumber \\&\quad \le - \frac{5}{2} \left( (x_0-\alpha (x')) \left( x_0 - \frac{1}{5} \alpha (x') - \frac{4}{5} \overline{x}_0 \right) \partial _{x_2} u, \partial _{x_2} u \right) _{\Omega _{k,h, \eta }} \nonumber \\&\qquad + ( \, ^t Pu, (x_0 - \overline{x}_0) \partial _{x_0} u) + ( (x_0 - \overline{x}_0) \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, (x_0 - \overline{x}_0) \partial _{x_0} u) \nonumber \\&\qquad - ( (x_0 - \overline{x}_0) \partial _{x_0} u, \, ^t (P-P_2)u), \end{aligned}$$
(11)

where \(\Omega _{\overline{x}_0, k', \eta } = \left\{ x \in \Omega _{\overline{x}_0,k'}: \ \eta \alpha (x') + (1- \eta ) \overline{x}_0 \le x_0 \le \alpha (x') \right\} \), with \(0 < \eta \le \frac{1}{5}\).

In \(\Omega _{\overline{x}_0,k', \eta }\), we consider the following inner products

$$\begin{aligned} (\, ^t Pu, \partial _{x_0} u) + ( \partial _{x_0} u, \, ^t Pu). \end{aligned}$$

Proceeding as done above, we obtain

$$\begin{aligned}&2 ( ( x_0 - \alpha (x')) \partial _{x_2} u, \partial _{x_2} u) - 4 ((x_0-\alpha (x')) \partial _{x_2} u, \partial _{x_0} u) \nonumber \\&\quad = - \int _{\Gamma } \left[ (\partial _{x_0} u)^2 + 2 \partial _{x_1} \alpha (x') \partial _{x_0} u \, \partial _{x_1} u + ( \partial _{x_1} u )^2 \right] {\text {d}} \sigma \nonumber \\&\qquad + \int _{\Gamma _{\eta , (1-\eta ) \bar{x}_{0}}} \left[ (\partial _{x_0} u)^2 + 2 \partial _{x_1} \alpha (x') \partial _{x_0} u \, \partial _{x_1} u + ( \partial _{x_1} u )^2 \right] {\text {d}} \sigma \nonumber \\&\qquad + ( \, ^t Pu, \partial _{x_0} u) + ( \partial _{x_0} u, \, ^t Pu) - ( \, ^t (P- P_2) u, \partial _{x_0} u) - ( \partial _{x_0} u, \, ^t (P-P_2)u), \end{aligned}$$
(12)

where \(\Gamma _{\eta , (1-\eta ) \overline{x}_0}\) is the surface \(x_0 = \eta \alpha (x') + (1- \eta ) \overline{x}_0\), with \(0< \eta \le \frac{1}{5}\). Making use of (11) and (12), we deduce the claim assuming that the gradient of u with respect to \(x_0\) and \(x_1\) is zero on \(\Gamma _{\eta , (1-\eta ) \overline{x}_0}\).

Let u be a function belonging to \(C_{0}^{\infty }(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0}\) and \(u|_S=0\). Let \(\chi \) be a function of class \(C^{\infty }\) such that \(\chi (t)=1\), if \(|t| \ge \eta \), and \(\chi (t)=0\), if \(|t| < \frac{\eta }{2}\). Rewriting (12) for \(u \chi \left( \dfrac{x_0}{\alpha (x')} \right) \) and adding (11), there exists \(0 < k \le k'\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le \varepsilon \Vert \, ^t Pu \Vert , \\&\qquad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0, k} \cap J_{k, \delta , \overline{x}_0}, \ u|_S=0. \end{aligned}$$

Finally, the claim follows from the previous inequality and by using Lemma 5 for k small enough. \(\square \)

5 Estimates under the assumption \(|\partial _{x_1} \alpha (x')| \ge 1\)

For every \(\overline{x}_0 \ge 0\), we set

$$\begin{aligned} \Gamma _{\overline{x}_0} = \{ x \in \Gamma : \ \alpha (x')= \overline{x}_0 \}. \end{aligned}$$

The next result holds.

Theorem 8

Let (i), (ii) and (iii) be satisfied. Let us assume that \(|\partial _{x_1}\alpha (x')| \ge 1\), on \(\Gamma _{\overline{x}_0}\). Then, there exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\ \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t Pu \Vert ,\nonumber \\&\forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k}= [\overline{x}_0, \overline{x}_0+k[ \times \Omega _0, \ u|_S=0, \end{aligned}$$
(13)

Moreover, for every \(\varepsilon > 0\) there exists \(k>0\) such that

$$\begin{aligned}&\Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t Pu \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) ,\nonumber \\&\qquad \, \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k}, \ u|_S=0. \end{aligned}$$
(14)

Proof

Let d > 0 and let us set

$$\begin{aligned} A_d = ( x_0+d) \partial _{x_0} + g_d(x') \partial _{x_1}, \end{aligned}$$

where \(g_d(x') = \dfrac{\alpha (x') +d}{\partial _{x_1} \alpha (x')}\), and consider the sum of the inner products

$$\begin{aligned} ( \, ^{t}Pu, A_d u) + (A_d u, \, ^{t} Pu)= & {} (\, ^{t}P_2u, A_d u) + (A_d u, \, ^{t}P_2u) +(\, ^{t}P_1u, A_d u) + (A_d u,\, ^{t} P_1u) \\&+ (\, ^{t}P_0u, A_d u) + (A_d u,\, ^{t} P_0u). \end{aligned}$$

For every \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(u|_S=0\), it results:

$$\begin{aligned}&(\, ^{t}P_2u, A_d u) + (A_d u, \, ^{t}P_2u) \nonumber \\&\quad = ( \, ^{t}P_2u, (x_0+d) \partial _{x_0} u) + ((x_0+d) \partial _{x_0} u,\ ^{t} P_2u) \nonumber \\&\quad \quad +(\, ^{t} P_2u, g_d(x') \partial _{x_1} u) + (g_d(x') \partial _{x_1} u, \, ^{t}P_2u). \end{aligned}$$
(15)

Let us integrate by parts in the first inner products of the principal part in (15)

$$\begin{aligned}&2(\, ^{t}P_2 u, (x_0+d) \partial _{x_0} u) \nonumber \\&\quad = (\partial _{x_0} u, \partial _{x_0} u) + (\partial _{x_1} u, \partial _{x_1} u) + \left( (x_0 - \alpha (x'))^2 \partial _{x_2} u, \partial _{x_2} u \right) \nonumber \\&\quad \quad + 2 \left( (x_0- \alpha (x')) (x_0+d) \partial _{x_2}u, \partial _{x_2}u \right) \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, (x_0+d) \partial _{x_0} u \right) \nonumber \\&\quad \quad + \int _{\Omega _0} (\overline{x}_0 +d) \left[ (\partial _{x_0} u)^2 + ( \partial _{x_1} u )^2 + (x_0 - \alpha (x'))^2 (\partial _{x_2} u )^2 \right] {\text {d}} x'. \end{aligned}$$
(16)

Moreover, integrating by parts in the second inner products in (15), we have

$$\begin{aligned}&2(\ ^{t}P_2u, g_d(x') \partial _{x_1} u) \nonumber \\&\quad = - (\partial _{x_0} u, \partial _{x_1} g_d(x') \partial _{x_0}u) \nonumber \\&\quad \quad + 4 ((x_0-\alpha (x')) \partial _{x_2} \alpha (x') g_d(x') \partial _{x_2} u, \partial _{x_1}u) \nonumber \\&\quad \quad - 2 \left( (x_0-\alpha (x')^2 \partial _{x_2} g_d(x') \partial _{x_2}u, \partial _{x_1} u \right) \nonumber \\&\quad \quad + \left( (x_0 -\alpha (x'))^2 \partial _{x_1} g_d(x') \partial _{x_2}u, \partial _{x_2}u \right) \nonumber \\&\quad \quad -2 \left( (x_0 - \alpha (x')) \partial _{x_1} \alpha (x') g_d(x') \partial _{x_2}u, \partial _{x_2} u \right) \nonumber \\&\quad \quad + \int _S n_1 g_d(x') (\partial _{x_0} u)^2 d \sigma + \int _S n_1 g_d(x') (\partial _{x_1} u)^2 {\text {d}} \sigma \nonumber \\&\quad \quad + 2 \int _S n_2 (x_0 - \alpha (x'))^2 g_d(x') \partial _{x_0} u \partial _{x_1} u {\text {d}} \sigma \nonumber \\&\quad \quad - \int _S n_1 (x_0 - \alpha (x'))^2 g_d(x') (\partial _{x_2} u)^2 {\text {d}} \sigma \nonumber \\&\quad \quad + \int _{\Omega _0} 2 g_d(x') \partial _{x_0} u \partial _{x_1} u {\text {d}} x'. \end{aligned}$$
(17)

Since \(u|_S=0\), it results

$$\begin{aligned} \int _{S} n_1 g_d(x') (\partial _{x_0} u)^2 {\text {d}} \sigma =0. \end{aligned}$$
(18)

Making use of the assumption (iii), it follows

$$\begin{aligned} \int _{S} n_1 g_d(x') (\partial _{x_1}u)^2 {\text {d}} \sigma \ge 0. \end{aligned}$$
(19)

Denoting the tangential derivative of u along the section of S of the equal height by \(\dfrac{\partial u}{\partial \tau }\), we obtain

$$\begin{aligned}&\ \ 2 \int _{S} n_2 (x_0- \alpha (x'))^2 g_d(x') \partial _{x_2} u \partial _{x_1} u d \sigma - \int _{S} n_1 (x_0-\alpha (x'))^2 g_d(x') (\partial _{x_2}u)^2 {\text {d}} \sigma \nonumber \\&\ \quad = - 2 \int _S (x_0- \alpha (x')) \left( \frac{\partial u}{\partial \tau } \right) g_d(x') \partial _{x_2} u d \sigma + \int _{S} n_1 (x_0-\alpha (x'))^2 g_d(x') (\partial _{x_2}u)^2 {\text {d}} \sigma \nonumber \\&\ \quad = \int _{S} n_1 (x_0-\alpha (x'))^2 g_d(x') (\partial _{x_2}u)^2 {\text {d}} \sigma \ge 0, \end{aligned}$$
(20)

where we took into account that \(\dfrac{\partial u}{\partial \tau }=0\), since \(u=0\) on S.

Adding (16) and (17) and making use of (18), (19) and (20), we have

$$\begin{aligned}&2 (\, ^{t}P_2u,A_d u) \nonumber \\&\quad \ge \Vert h^{\frac{1}{2}}(x') \partial _{x_0} u \Vert ^2 + \Vert h^{\frac{1}{2}}(x') \partial _{x_1} u \Vert ^2 + \Vert (4- h(x'))^{\frac{1}{2}} \left( x_0-\alpha (x')\right) \partial _{x_2} u \Vert ^2 \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') (x_0+d) \partial _{x_2} u, \partial _{x_0}u\right) \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') g_d(x') \partial _{x_2} u, \partial _{x_1} u \right) \nonumber \\&\quad \quad - 2 \left( (x_0-\alpha (x'))^2 \partial _{x_2} g_d(x') \partial _{x_2} u, \partial _{x_1} u \right) \nonumber \\&\quad \quad + \int _{\Omega _0} \left\{ (\overline{x}_0 +d) \left[ (\partial _{x_0} u)^2 + ( \partial _{x_1} u )^2 + (x_0 - \alpha (x'))^2 (\partial _{x_2} u )^2 \right] + 2 g_d(x') \partial _{x_0} u \partial _{x_1} u \right\} {\text {d}} x' \nonumber \\&\quad = \Vert h^{\frac{1}{2}}(x') \partial _{x_0} u \Vert ^2 + \Vert h^{\frac{1}{2}}(x') \partial _{x_1} u \Vert ^2 + \Vert (4- h(x'))^{\frac{1}{2}} \left( x_0-\alpha (x')\right) \partial _{x_2} u \Vert ^2 \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x'))^2 \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x')) \alpha (x') \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad + 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') g_d(x') \partial _{x_2} u, \partial _{x_1} u \right) \nonumber \\&\quad \quad - 2 \left( (x_0-\alpha (x'))^2 \partial _{x_2} g_d(x') \partial _{x_2} u, \partial _{x_1} u \right) \nonumber \\&\quad \quad + \int _{\Omega _0} \Big \{ (\overline{x}_0 +d) \left[ (\partial _{x_0} u)^2 + ( \partial _{x_1} u )^2 + (x_0 - \alpha (x'))^2 (\partial _{x_2} u )^2 \right] + 2 g_d(x') \partial _{x_0} u \partial _{x_1} u \Big \} {\text {d}} x'. \end{aligned}$$
(21)

By assumption (i), there exist two positive numbers k and \(\delta \) such that, for \(d > \frac{1}{h_1} |g(x')|\), where \(x' \in \Omega _0 \cap J_{k, \delta , \overline{x}_0}\), it results \((\overline{x}_0 +d)^2 - (g_d(x'))^2 \ge 0\) and, hence,

$$\begin{aligned} \int _{\Omega _0} \left[ (\overline{x}_0 +d) (\partial _{x_0} u)^2 + 2 g_d(x') \partial _{x_0} u \partial _{x_1} u + (\overline{x}_0 +d) (\partial _{x_1} u )^2 \right] {\text {d}} x' \ge 0. \end{aligned}$$

By using (21), we deduce

$$\begin{aligned}&(\ ^{t}P_2 u, A_du) + (A_du, \ ^{t}P_2 u) \nonumber \\&\quad = 2 (\ ^{t}P_2 u, A_du) \nonumber \\&\quad \ge \Vert h^{\frac{1}{2}}(x') \partial _{x_0} u \Vert ^2 + \Vert h^{\frac{1}{2}}(x') \partial _{x_1} u \Vert ^2 + \Vert [4- h(x')]^{\frac{1}{2}} (x_0-\alpha (x')) \partial _{x_2} u \Vert ^2 \nonumber \\&\qquad + 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\qquad -2 \left( (x_0-\alpha (x'))^2 \partial _{x_2} g_d (x') \partial _{x_2} u, \partial _{x_1}u \right) \nonumber \\&\qquad + 4 \left( (x_0-\alpha (x'))^2 \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\qquad + 4 \left( (x_0-\alpha (x')) \alpha (x') \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) . \end{aligned}$$
(22)

Now, we consider the first-order terms. Integrating by parts, it results

$$\begin{aligned}&(^{t}P_1u, A_du) + (A_du, \ ^{t}P_1u) \nonumber \\&\quad = -8 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, (x_0+d) \partial _{x_0}u+ g_d(x') \partial _{x_1} u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_0(x) \partial _{x_0} u + \widetilde{a}_1(x) \partial _{x_1} u + (x_0 - \alpha (x')) \widetilde{a}_2(x) \partial _{x_2} u, (x_0+d) \partial _{x_0} u + g_d(x') \partial _{x_1} u \right) \nonumber \\&\quad = -8 \left( (x_0-\alpha (x'))^2 \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad -8 \left( (x_0-\alpha (x')) \alpha (x') \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad -8 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_0(x) (x_0-\alpha (x')) \partial _{x_0} u, (x_0+d) \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_0(x) \alpha (x') \partial _{x_0} u, (x_0+d) \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_1(x) (x_0-\alpha (x')) \partial _{x_1} u, (x_0+d) \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_1(x) \alpha (x') \partial _{x_0} u, (x_0+d) \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_2(x) (x_0-\alpha (x'))^2 \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_2(x) (x_0- \alpha (x')) \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_0(x) \partial _{x_0} u, g_d(x') \partial _{x_1} u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_1(x) \partial _{x_1} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_2(x) (x_0-\alpha (x')) \partial _{x_2} u, g_d(x') \partial _{x_1}u \right) \end{aligned}$$
(23)

Adding (22) and (23), we have

$$\begin{aligned}&(\ ^{t}P u, A_du) + (A_du, \ ^{t}P u) \nonumber \\&\quad \ge h_1 \Vert \partial _{x_0} u \Vert ^2 + h_1 \Vert \partial _{x_1} u \Vert ^2 + (4-h_2) \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert ^2 \nonumber \\&\quad \quad - 4 \left( (x_0-\alpha (x')) \partial _{x_2} \alpha (x') \partial _{x_2} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\quad \quad - 4 \left( (x_0-\alpha (x'))^2 \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad - 4 \left( (x_0-\alpha (x')) \alpha (x') \partial _{x_2} \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left| \left( (x_0-\alpha (x'))^2 \partial _{x_2} g_d(x') \partial _{x_2} u, \partial _{x_1}u \right) \right| \nonumber \\&\quad \quad -c \Vert (x_0+d) \partial _{x_0} u \Vert \Vert \partial _{x_0} u \Vert \nonumber \\&\quad \quad -2 \left( \widetilde{a}_0(x) (x_0-\alpha (x')) \partial _{x_0} u, x_0 \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_0(x) \alpha (x') \partial _{x_0} u, x_0 \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_1(x) (x_0-\alpha (x')) \partial _{x_1} u, (x_0+d) \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_1(x) \alpha (x') \partial _{x_0} u, x_0 \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_2(x) (x_0-\alpha (x'))^2 \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_2(x) (x_0- \alpha (x')) \alpha (x') \partial _{x_2} u, \partial _{x_0}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_0(x) \partial _{x_0} u, g_d(x') \partial _{x_1} u \right) \nonumber \\&\quad \quad - 2 \left( \widetilde{a}_1(x) \partial _{x_1} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\quad \quad -2 \left( \widetilde{a}_2(x) (x_0-\alpha (x')) \partial _{x_2} u, g_d(x') \partial _{x_1}u \right) \nonumber \\&\quad \quad - \left| (\, ^t P_0u, A_du) \right| - \left| (A_du, \, ^t P_0u) \right| , \quad \forall u \in C^{*\infty }_0(\overline{\Omega }_k). \end{aligned}$$
(24)

Since \(\alpha (x')\), \(g_d(x')\) and \(\beta (x')\) vanish on \(\Omega _0 \cap \Gamma \), for every \(\delta >0\) there exist a neighborhood \(I_{x'}\) in \(\Omega _0 \cap \Gamma \) and \(k>0\) such that

$$\begin{aligned}&|\alpha (x')|< \delta , \ | g_d(x')|< \delta , \quad \forall x' \in I_{x'},\\&|x_0 - \alpha (x')| < \delta , \quad \forall x \in [0,k[ \times I_{x'}. \end{aligned}$$

Let \(\varphi \in C^{\infty }_0(\overline{\Omega })\) such that \(\varphi \equiv 1\), on \([0,k'[ \times I'_{x'}\), with \(I'_{x'} \subseteq I_{x'}\) and \(k'<k\), \(0 \le \varphi (x) \le 1\) and \(\mathrm{supp} \, \varphi \subseteq [0, k[ \times I_{x'}\). Without lost generality, we can consider \([0, k'[ \times I'_{x'}\) such that \(|x_0 - \alpha (x')| \ge \dfrac{\varepsilon }{2}\), for every \(x \in \Omega _k \setminus ([0, k'[ \times I'_{x'})\). Using (22) and the previous remarks, it follows

$$\begin{aligned}&h_1 \Vert \partial _{x_0} u \Vert ^2 + h_1 \Vert \partial _{x_1} u \Vert ^2 + (4- h_2) \Vert (x_0- \alpha (x') \partial _{x_2} u \Vert ^2 \\&\quad \le c (\delta +k) ( \Vert \varphi ^{\frac{1}{2}}(x) \partial _{x_0} u \Vert ^2 + \Vert \varphi ^{\frac{1}{2}}(x) \partial _{x_1} u \Vert ^2 + \Vert \varphi ^{\frac{1}{2}}(x) (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) \\&\qquad + c ( \Vert (1-\varphi (x))^{\frac{1}{2}} \partial _{x_0} u \Vert ^2 + \Vert (1-\varphi (x))^{\frac{1}{2}} \partial _{x_1} u \Vert ^2 + \Vert (1-\varphi (x))^{\frac{1}{2}} (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) \\&\qquad + 2 (\Vert \, ^t P_0u \Vert \Vert A_du \Vert + \Vert \, ^t Pu \Vert \Vert A_d u \Vert ). \end{aligned}$$

Taking into account Lemma 1, we get

$$\begin{aligned}&h_1 \Vert \partial _{x_0} u \Vert ^2 + h_1 \Vert \partial _{x_1} u \Vert ^2 + (4- h_2) \Vert (x_0- \alpha (x') \partial _{x_2} u \Vert ^2 \\&\quad \le c (\delta +k) ( \Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) + c ( \Vert (1-\varphi (x))^{\frac{1}{2}} \partial _{x_0} u \Vert ^2 \\&\qquad + \Vert (1-\varphi (x))^{\frac{1}{2}} \partial _{x_1} u \Vert ^2 + \Vert (1-\varphi (x))^{\frac{1}{2}} (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) + c (\Vert \, ^t Pu \Vert + \Vert \, ^t P_0 u\Vert ) \\&\quad = c (\delta +k) ( \Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) \\&\qquad + c ( \Vert \partial _{x_0} (1-\varphi (x))^{\frac{1}{2}} u + [(1-\varphi )^{\frac{1}{2}}, \partial _{x_0}] u \Vert ^2 + \Vert \partial _{x_1} (1-\varphi (x))^{\frac{1}{2}} u + [(1-\varphi )^{\frac{1}{2}}, \partial _{x_1}] u \Vert ^2 \\&\qquad + \Vert (x_0 - \alpha (x')) \partial _{x_2} (1-\varphi (x))^{\frac{1}{2}} u + (x_0 - \alpha (x')) [(1-\varphi )^{\frac{1}{2}}, \partial _{x_2}] u \Vert ^2 ) + c (\Vert \, ^t Pu \Vert + \Vert \, ^t P_0 u\Vert ) \\&\quad \le c (\delta +k) ( \Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 ) + c ( \Vert \partial _{x_0} (1-\varphi (x))^{\frac{1}{2}} u \Vert ^2 \\&\qquad + \Vert \partial _{x_1} (1-\varphi (x))^{\frac{1}{2}} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} (1-\varphi (x))^{\frac{1}{2}} u \Vert ^2 ) + c (\Vert \, ^t Pu \Vert + \Vert \, ^t P_0 u\Vert ). \end{aligned}$$

Making use of Lemmas 1 and 3, for \(\delta \) and k small enough, we obtain

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0- \alpha (x') \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t Pu \Vert , \nonumber \\&\qquad \qquad \quad \forall u \in C_0^{\infty } (I_{\overline{x}_0,k, \delta }): \ u|_S=0. \end{aligned}$$
(25)

Let \(\chi \in C_0^{\infty }(\mathbb {R})\) such that \(\chi (t)=1\), if \(|t| \le \frac{1}{2}\), and \(\chi (t)=0\), if \(|t| >1\). We rewrite (25) for \(u \chi \left( \frac{x_0 - \alpha (x')}{\delta } \right) \) and apply Lemma 5 to \(u \left[ 1- \chi \left( \frac{x_0 - \alpha (x')}{\delta } \right) \right] \). Adding the obtained estimates, for \(\delta \) small enough and k suitable and small, we reach (13).

Instead, in order to get (14), let γ > 0 and let us consider the operator

$$\begin{aligned} A_{\overline{x}_0, \gamma }= (x_0 - \alpha (x')) \partial _{x_0} + g_{\overline{x}_0, \gamma }(x') \partial _{x_1}, \end{aligned}$$

where

$$\begin{aligned} g_{\overline{x}_0, \gamma }(x') = \frac{\alpha (x') - \overline{x}_0 + \gamma }{\partial _{x_1} \alpha (x')}. \end{aligned}$$

Integrating by parts in the inner products \((\, ^t Pu, A_{\overline{x}_0, \gamma }u) + (A_{\overline{x}_0, \gamma }u, \, ^t Pu)\), using the same arguments as done and since \(g_{\overline{x}_0, \gamma }(x')\) has the same sign of \(g(x')\) on \(S \cap I_{k, \delta }\), we deduce

$$\begin{aligned}&\ \ (4- \varepsilon ) \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 + \int _{\Omega _0} \left[ \gamma (\partial _{x_0} u)^2 + 2 g_{\overline{x}_0, \gamma }(x') \partial _{x_0} u \, \partial _{x_1} u + \gamma ( \partial _{x_1} u )^2 \right] {\text {d}} x' \nonumber \\&\ \quad \le c \varepsilon ( \Vert \partial _{x_0} u \Vert ^2 + \Vert \partial _{x_1} u \Vert ^2 + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert ^2 + \Vert u \Vert ^2 + \Vert \, ^t Pu \Vert ^2), \nonumber \\&\ \quad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k} \cap I_{k, \delta }, \ u|_S=0. \end{aligned}$$
(26)

For \(\delta \) small enough and since \(|\partial _{x_1} \alpha (x')| >1\), on \(\Gamma _{\overline{x}_0}\), it results

$$\begin{aligned} \int _{\Omega _0} \left[ \gamma (\partial _{x_0} u)^2 + 2 g_{\overline{x}_0, \gamma }(x') \partial _{x_0} u \, \partial _{x_1} u + \gamma ( \partial _{x_1} u )^2 \right] {\text {d}} x' >0. \end{aligned}$$

As a consequence, for \(\varepsilon \) small enough and k suitable and small, we have

$$\begin{aligned}&\Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t Pu \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) , \\&\quad \qquad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k} \cap I_{\Gamma _{\overline{x}_0}}, \ u|_S=0. \end{aligned}$$

Rewriting the previous inequality for \(u \chi \left( \frac{x_0 - \alpha (x')}{\delta } \right) \) and applying Lemma 6 to \(u \left[ 1- \chi \left( \frac{x_0 - \alpha (x')}{\delta } \right) \right] \), as done above, (14) follows for \(\overline{x}_0 >0\).

On the other hand, if \(\overline{x}_0 =0\), considering the inner products

$$\begin{aligned} (A_{\overline{x}_0, \gamma } u, \, ^t Pu) + ( \, ^t Pu, A_{\overline{x}_0, \gamma } u) \end{aligned}$$

and proceeding as before, we obtain (13) and, then, (14) for \(\gamma \) small enough. \(\square \)

6 Conclusive a priori estimates

Let us assume that \(|\partial _{x_1} \alpha (x') |=1\) in some points of the plane \(x_0 = \overline{x}_0\), with \(\overline{x}_0 > 0\). Let \(\Omega _{\overline{x}_0}\) be the intersection between the plane \(x_0 = \overline{x}_0\) and \(\Omega \). Let \(\Gamma _{\overline{x}_0} = \Gamma \cap \Omega _{\overline{x}_0}\). Let \(\Gamma _{\overline{x}_0}'\) be the set of the points of \(\Gamma _{\overline{x}_0}\) where \(\partial _{x_1} \alpha (x_1)=1\) and, finally, let \(I_{\overline{x}_0}\) be a neighborhood of \(\overline{x}_0\) in \(\Gamma _{\overline{x}_0}'\) on \(\Omega _{\overline{x}_0}\) such that \(\partial _{x_1} \alpha (x_1) \lessgtr 1\) outside \(I_{\overline{x}_0}\). The following result holds.

Theorem 9

Let (i), (ii) and (iii) be satisfied. If on the plane \(x_0 = \overline{x}_0>0\) there exist points in which \(|\partial _{x_1} \alpha (x')|=1\), then there exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t Pu \Vert , \nonumber \\&\qquad \quad \ \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0, k}, \ u|_S=0. \end{aligned}$$
(27)

Moreover, for every \(\varepsilon >0\) there exists \(k>0\) such that

$$\begin{aligned}&\Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t Pu \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) ,\nonumber \\&\qquad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0, k}, \ u|_S=0. \end{aligned}$$
(28)

Proof

Let \(\Omega _{\overline{x}_0} \cap \Gamma \), let \(\overline{x}' \in \Omega _0\) such that \(| \partial _{x_1} \alpha (\overline{x}')|=1\). We set

$$\begin{aligned} \gamma (\overline{x}') = {\left\{ \begin{array}{ll} \begin{array}{ll} \sqrt{\overline{x}_0} - \sqrt{\alpha (\overline{x}')}, &{} \quad \mathrm{if} \ \partial _{x_1} \alpha (\overline{x}') =1, \\ -(\sqrt{\overline{x}_0} - \sqrt{\alpha (\overline{x}')}), &{} \quad \mathrm{if} \ \partial _{x_1} \alpha (\overline{x}') =-1. \end{array} \end{array}\right. } \end{aligned}$$

Evidently, it results \(|\partial _{x_1} \alpha (x')| \le 1\) on the curve \(x_0 - \overline{x}_0 = \gamma (x')\) and \(x \in J_{k, \delta , \overline{x}_0}\), with suitable k and \(\delta \). Therefore, there exists \(\eta \) such that \(|\partial _{x_1} \alpha (x')| \le 1\) if \(|x_0 - \overline{x}_0| \le \eta \gamma (x')\) and \(x \in J_{k, \delta , \overline{x}_0}\). Whereas \(|\partial _{x_1} \alpha (x')| \ge 1\) on \(\Omega _{\overline{x}_0}\) if \(|x_0 - \overline{x}_0| \ge \eta \gamma (x')\). Let \(\chi \in C^{\infty }(\mathbb {R})\) such that \(\chi (t)=0\) if \(t \le \frac{\eta }{2}\) and \(\chi (t) =1\) if \(t \ge \eta \). For every \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq J_{k, \delta , \overline{x}_0}\) and \(u|_S=0\), we rewrite (13) and (14) for \(\chi \left( \frac{x_0 - \overline{x}_0}{\gamma (x')} \right) u\) and (10) for \(\left( 1- \chi \left( \frac{x_0 - \overline{x}_0}{\gamma (x')} \right) \right) u\). Adding such inequalities, for k small enough, we have

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert ( x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t P u \Vert , \nonumber \\&\qquad \ \, \forall u \in C^{\infty }_0(\Omega _{k, \overline{x}_0}): \ \mathrm{supp} \, u \subseteq J_{k, \delta , \overline{x}_0}, \ u|_S=0, \end{aligned}$$
(29)

and

$$\begin{aligned}&\Vert ( x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t P u \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) ,\nonumber \\&\qquad \quad \forall u \in C^{\infty }_0(\Omega _{k, \overline{x}_0}): \ \mathrm{supp} \, u \subseteq J_{k, \delta , \overline{x}_0}, \ u|_S=0. \end{aligned}$$
(30)

From (29), (30) and Lemma 5, it follows

$$\begin{aligned}&\ \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert ( x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t P u \Vert , \\&\forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \overline{\Omega }_{\overline{x}_0, k} = [\overline{x}_0, \overline{x}_0 + k[ \times \overline{\Omega }_0, \ u|_S=0, \end{aligned}$$

and

$$\begin{aligned}&\Vert ( x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t P u \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) , \\&\forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \overline{\Omega }_{\overline{x}_0, k} = [\overline{x}_0, \overline{x}_0 + k[ \times \overline{\Omega }_0, \ u|_S=0. \end{aligned}$$

\(\square \)

With the same techniques used in Theorem 6 if \(\overline{x}_0 =0\) and Theorems 8 and 9 if \(\overline{x}_0 >0\), we obtain the next result.

Theorem 10

Let (i), (ii) and (iii) be satisfied. If on the plane \(x_0 = \overline{x}_0>0\) there exist points in which \(|\partial _{x_1} \alpha (x')|=1\), then there exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t Pu \Vert , \nonumber \\&\quad \ \forall u \in C_{0}^{\infty }(\widetilde{\Omega }): \ \mathrm{supp} \, u \subseteq \widetilde{\Omega }_{\overline{x}_0, k} = [\overline{x}_0, \overline{x}_0 + k[ \times \widetilde{\Omega }_0. \end{aligned}$$
(31)

Moreover, for every \(\varepsilon >0\) there exists \(k>0\) such that

$$\begin{aligned}&\Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert \le \varepsilon \left( \Vert \, ^t Pu \Vert + \Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert u \Vert \right) ,\nonumber \\&\qquad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \widetilde{\Omega }_{\overline{x}_0, k}= [\overline{x}_0, \overline{x}_0 + k[ \times \widetilde{\Omega }_0. \end{aligned}$$
(32)

7 Estimates in Sobolev spaces with \(s<0\)

Let \(\Omega _{\overline{x}_0}'\) be the intersection between \(\Omega '\) and the plane \(x_0 = \overline{x}_0\). Let \(\Gamma _{\overline{x}_0}'\) be the set of points belonging into \(\Gamma _{\overline{x}_0}= \Gamma \cap \Omega _{\overline{x}_0}'\) such that \(|\partial _{x_1} \alpha (x') |=1\). Moreover, let \(J_{\overline{x}_0}\) be the intersection between a neighborhood of \(\Gamma _{\overline{x}_0}'\) and \(\Omega _{\overline{x}_0}'\). We are able to prove the following estimate in Sobolev spaces with \(s<0\).

Theorem 11

Let (i), (ii) and (iii) be satisfied. Then, for every \(\overline{x}_0 \ge 0\) and for every \(s<0\) there exist \(k>0\) and \(c>0\) such that

$$\begin{aligned}&\quad \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0, s}} + \Vert (x_0 -\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert Pu \Vert _{H^{0,s}}, \nonumber \\&\ \qquad \quad \forall u \in C^{\infty }_0(\overline{\Omega }_k): \ \mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k} = [\overline{x}_0, \overline{x}_0+k[ \times \Omega _0. \end{aligned}$$
(33)

Proof

Firstly, let \(x_0>0\). Let \(\varphi \in C^{\infty }_0(\mathbb {R}^2)\) such that \(\mathrm{supp} \, \varphi \subseteq \Omega _{0}'\), \(\varphi \equiv 1\) on \(\Omega _0\), with \(\Omega _0 \subset \Omega _0'\). For every \(u \in C^{\infty }_0(\overline{\Omega }_k)\) such that \(\mathrm{supp} \, u \subseteq \Omega _{\overline{x}_0,k}= [\overline{x}_0, \overline{x}_0+k[ \times \Omega _0\), we set \(v_s= \varphi (x') A_s u\). Making use of Theorem 10, it follows

$$\begin{aligned} \Vert \partial _{x_0} v_s \Vert + \Vert \partial _{x_1} v_s \Vert + \Vert (x_0 -\alpha (x')) \partial _{x_2} v_s \Vert + \Vert v_s \Vert \le c \Vert \, ^t P v_s \Vert . \end{aligned}$$
(34)

We have

$$\begin{aligned} \Vert \partial _{x_0} v_s \Vert= & {} \Vert \partial _{x_0} \varphi (x') A_s u \Vert \nonumber \\= & {} \Vert \varphi (x') A_s \partial _{x_0} u \Vert \nonumber \\= & {} \Vert A_s \varphi (x') \partial _{x_0} u + [ \varphi , A_s] \partial _{x_0} u \Vert \nonumber \\\ge & {} \Vert A_s \partial _{x_0} u \Vert - \Vert R \partial _{x_0} u \Vert , \end{aligned}$$
(35)

where \(R= [ \varphi , A_s] u\) is a regularizing pseudodifferential operator.

By using (35) and Lemma 4, we obtain

$$\begin{aligned} \Vert \partial _{x_0} v_s \Vert\ge & {} \Vert A_s \partial _{x_0} u \Vert - c \Vert R (x_0- \overline{x}_0) \partial ^2_{x_0} u \Vert \nonumber \\= & {} \Vert A_s \partial _{x_0} u \Vert - c \Vert R (x_0- \overline{x}_0) (- \, ^t Pu + \, ^t Pu + \partial _{x_0}^2 u ) \Vert \nonumber \\\ge & {} \Vert A_s \partial _{x_0} u \Vert - c \Vert R (x_0- \overline{x}_0) \, ^t Pu \Vert - c \Vert R (x_0- \overline{x}_0) ( \, ^t Pu + \partial _{x_0}^2 u ) \Vert \nonumber \\\ge & {} \Vert \partial _{x_0} u \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) \, ^t Pu \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}} \nonumber \\&- c \Vert (x_0- \overline{x}_0) u \Vert _{H^{0,s}}. \end{aligned}$$
(36)

Furthermore, it results

$$\begin{aligned} \Vert \partial _{x_1} v_s \Vert= & {} \Vert \partial _{x_1} \varphi (x') A_s u \Vert \nonumber \\= & {} \Vert (\partial _{x_1} \varphi (x')) A_s u + \varphi (x') A_s \partial _{x_1} u \Vert \nonumber \\= & {} \Vert ( \partial _{x_1} \varphi (x')) A_s u + A_s \partial _{x_1} u + [ \varphi , A_s ] \partial _{x_1} u \Vert \nonumber \\\ge & {} \Vert A_s \partial _{x_1} u \Vert - \Vert R_1 A_s u \Vert - \Vert [ \varphi , A_s ] \partial _{x_1} u \Vert \nonumber \\\ge & {} \Vert \partial _{x_1} u \Vert _{H^{0,s}} - c \Vert u \Vert _{H^{0,s}} - \Vert R_2 \partial _{x_1} u \Vert \nonumber \\\ge & {} \Vert \partial _{x_1} u \Vert _{H^{0,s}} - c \Vert u \Vert _{H^{0,s}}, \end{aligned}$$
(37)

where \(R_1\) and \(R_2\) are regularizing pseudodifferential operators.

Finally, we get

$$\begin{aligned} \Vert (x_0 - \alpha (x')) \partial _{x_2} v_s \Vert= & {} \Vert (x_0 - \alpha (x')) \partial _{x_2} (\varphi (x') A_s u) \Vert \nonumber \\= & {} \Vert (\partial _{x_2} \varphi (x')) (x_0 - \alpha (x')) A_s u + (x_0 - \alpha (x')) \varphi (x') A_s \partial _{x_2} u \Vert \nonumber \\= & {} \Vert (\partial _{x_2} \varphi (x')) (x_0 - \alpha (x')) A_s u + (x_0 - \alpha (x')) A_s \varphi (x') \partial _{x_2} u \nonumber \\&+ (x_0 - \alpha (x')) [ \varphi , A_s] \partial _{x_2} u \Vert \nonumber \\= & {} \Vert R_3 u + A_s (x_0 - \alpha (x')) \varphi (x') \partial _{x_2} u + [x_0 - \alpha (x'), A_s] \varphi (x') \partial _{x_2} u \nonumber \\&+ R_4 \partial _{x_2} u \Vert \nonumber \\&\ge \Vert A_s (x_0 - \alpha (x')) \varphi (x') \partial _{x_2} u \Vert - \Vert R_3 u \Vert - \Vert R_4 \partial _{x_2} u \Vert - \Vert B_{s-1} \partial _{x_2} u \Vert \nonumber \\&\ge \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} - c \Vert u \Vert _{H^{0,s}} - \Vert B_s' u \Vert \nonumber \\&\ge \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} - c \Vert u \Vert _{H^{0,s}}, \end{aligned}$$
(38)

where \(R_3\) and \(R_4\) are regularizing pseudodifferential operators, \(B_{s-1}\) and \(B'_{s}\) are pseudodifferential operators of order \(s-1\) and s, respectively. Adding (36), (37), (38) and using Lemma 4, it follows

$$\begin{aligned}&\Vert \partial _{x_0} v_s \Vert + \Vert \partial _{x_1} v_s \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} v_s \Vert \nonumber \\&\quad \ge \Vert \partial _{x_0} u \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) P u \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) \partial _{x_0} u \Vert -c \Vert (x_0- \overline{x}_0) u \Vert _{H^{0,s}} \nonumber \\&\qquad + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} - c \Vert u \Vert _{H^{0,s}} \nonumber \\&\quad \ge \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) \, ^t Pu \Vert _{H^{0,s}} \nonumber \\&\qquad - c \Vert (x_0- \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}} - c \Vert (x_0- \overline{x}_0) u \Vert _{H^{0,s}}. \end{aligned}$$
(39)

Moreover, it results

$$\begin{aligned} \Vert \, ^t Pv_s \Vert= & {} \Vert \, ^t P(\varphi (x') A_s u) \Vert \nonumber \\= & {} \Vert \varphi (x') \, ^t P A_s u + [\varphi (x'), \, ^t P] A_s u \Vert \nonumber \\= & {} \Vert \varphi (x') A_s \, ^t P u + \varphi (x') [\, ^t P, A_s] u + R_5 u \Vert \nonumber \\= & {} \Vert A_s \, ^t P u + [\varphi (x'), A_s] \, ^t P u + \varphi (x') [\, ^t P, A_s] u + R_5 u \Vert \nonumber \\= & {} \Vert A_s \, ^t P u + R_6 \, ^t P u + \varphi (x') [\, ^t P, A_s] u + R_5 u \Vert , \end{aligned}$$
(40)

where \(R_5\) and \(R_6\) are regularizing operators.

The commutator \([\, ^t P, A_s]\) is given by

$$\begin{aligned} \varphi (x') [\, ^t P, A_s] u = \varphi (x') [\, ^t P_2, A_s] u + \varphi (x') [\, ^t P_1, A_s] u + \varphi (x') [\, ^t P_0, A_s] u. \end{aligned}$$
(41)

We consider the principal part:

$$\begin{aligned}{}[\, ^t P_2, A_s] u = B_{s+1} u + B_s u, \end{aligned}$$

where \(B_{s+1}\) and \(B_s\) are pseudodifferential operators of order \(s+1\) and s, respectively. The symbol of \(B_{s+1}\) is given by

$$\begin{aligned} b(x, \xi ')= & {} - \frac{1}{i} \sum _{h=1}^2 \partial _{x_h} (\xi ^2_1 +(x_0-\alpha (x'))^2 \xi ^2_2) \varphi (x') \partial _{\xi _h}(1+ |\xi '|^2)^{\frac{s}{2}} \\= & {} - \frac{1}{i} \left( 2 (x_0-\alpha (x')) (- \partial _{x_1} \alpha (x')) \xi ^2_2 \right) \varphi (x') \partial _{\xi _1}(1+ |\xi '|^2)^{\frac{s}{2}} \\&- \frac{1}{i} \left( 2 (x_0-\alpha (x')) (- \partial _{x_2} \alpha (x')) \xi ^2_2 \right) \varphi (x') \partial _{\xi _2}(1+ |\xi '|^2)^{\frac{s}{2}} \end{aligned}$$

Then, \(B_{s+1} u = (x_0-\alpha (x'))\varphi (x') \partial _{x_2} B_s'u\), where \(B_s'\) is a pseudodifferential operator of order s. Moreover, taking into account Theorem 10, we deduce

$$\begin{aligned} \Vert B_{s+1} u \Vert= & {} \Vert (x_0- \alpha (x')) \varphi (x') \partial _{x_2} B'_su \Vert \\\le & {} \varepsilon \left( \Vert \, ^t P B_s' u \Vert + \Vert \partial _{x_0} B_s'u \Vert + \Vert \partial _{x_1} B'_s u \Vert + \Vert B'_s u \Vert \right) \\\le & {} \varepsilon \left( \Vert B_s' \, ^t P u \Vert + \Vert [\, ^t P, B'_s] u \Vert + \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \right) \\\le & {} \varepsilon \big ( \Vert \, ^t P u \Vert _{H^{0,s}} + \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \big ), \end{aligned}$$

being \([\, ^t P, B'_s]\) a pseudodifferential operator of order \(s-1\) and its principal symbol \(b'(x, \xi )\) of the same type of \(b(x, \xi )\). Hence, making use of Lemma 4, it results

$$\begin{aligned} \Vert \varphi (x') [^t P_2, A_s] u \Vert\le & {} \varepsilon c \big ( \Vert \, ^t Pu \Vert _{H^{0,s}} + \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \nonumber \\&+ \Vert u \Vert _{H^{0,s}} \big ) + \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}}. \end{aligned}$$
(42)

We consider the first-order part of the commutator

$$\begin{aligned} \varphi (x') [\, ^t P_1, A_s] u = B_{s-1} \partial _{x_0} u + B_{s} u + B_{s-1} u, \end{aligned}$$
(43)

where \(B_{s-1}\) and \(B_{s}\) are pseudodifferential operators of order \(s-1\) and s, respectively.

By using Lemma 4, we have

$$\begin{aligned} \Vert B_{s-1} \partial _{x_0} u \Vert\le & {} c \Vert (x_0 - \overline{x}_0) \partial _{x_0} B_{s-1} \partial _{x_0} u \Vert \\= & {} c \Vert (x_0 - \overline{x}_0) B_{s-1} \partial ^2_{x_0} u \Vert \\\le & {} c \big ( \Vert (x_0 - \overline{x}_0) B_{s-1} P u \Vert + \Vert (x_0 - \overline{x}_0) B_{s}' \partial _{x_0} u \Vert \\&+ \Vert (x_0 - \overline{x}_0) B_{s}'' (x_0 - \alpha (x')) \partial _{x_1} u \Vert \\&+ \Vert (x_0 - \overline{x}_0) B_{s}''' (x_0 - \alpha (x')) \partial _{x_2} u \Vert \\&+ \Vert (x_0 - \overline{x}_0) B_s^{(iv)} u \Vert \big ), \end{aligned}$$

where \(B^{(i)}_s\) are pseudodifferential operators of order s. Hence, it results

$$\begin{aligned} \ \ \Vert B_{s-1} \partial _{x_0} u \Vert\le & {} c \big ( \Vert (x_0 - \overline{x}_0) P u \Vert _{H^{0,s}} + \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}} + \Vert (x_0 - \overline{x}_0) \partial _{x_1} u \Vert _{H^{0,s}} \nonumber \\&\ + \Vert (x_0 - \overline{x}_0) (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert (x_0- \overline{x}_0) u \Vert _{H^{0,s}} \big ). \end{aligned}$$
(44)

Taking into account (43) and (44), it follows

$$\begin{aligned} \Vert \varphi (x') [\, ^t P_1, A_s ] \Vert\le & {} c \big ( \Vert (x_0 - \overline{x}_0) P u \Vert _{H^{0,s}} + \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}} \nonumber \\&+ \Vert (x_0 - \overline{x}_0) (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \nonumber \\&+ \Vert (x_0 - \overline{x}_0) u \Vert _{H^{0,s}} + \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}} \big ). \end{aligned}$$
(45)

We estimate the zero-order part:

$$\begin{aligned} \Vert \varphi (x') [\, ^t P_0, A_s ] u \Vert\le & {} c \Vert u \Vert _{H^{0,s}} \nonumber \\\le & {} c \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}}. \end{aligned}$$
(46)

Making use of (42), (45), (46) and for \(| x_0 - \overline{x}_0 | \le k < \varepsilon \), we obtain

$$\begin{aligned} \Vert \varphi (x') [\, ^t P, A_s ] u \Vert\le & {} c \varepsilon \big ( \Vert \, ^t P u \Vert _{H^{0,s}} + \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} \nonumber \\&+ \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \big ). \end{aligned}$$
(47)

Taking into account (40), (47) and Lemma 4, denoted the generic regularizing operator by R, it follows

$$\begin{aligned} \Vert \, ^t Pv_s \Vert\le & {} \Vert A_s \, ^t Pu \Vert + \Vert R \, ^t Pu \Vert + \Vert \varphi (x') [P, A_s] u \Vert + \Vert R u \Vert \nonumber \\\le & {} c \Vert \, ^t P u \Vert _{H^{0,s}} + \varepsilon c \left( \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0 - \alpha (x')) \partial _{x_1} u \Vert _{H^{0,s}} \right) \nonumber \\&\quad + c \Vert u \Vert _{H^{0,s}} \nonumber \\\le & {} c \Vert \, ^t Pu \Vert _{H^{0,s}} + \varepsilon c \left( \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \right) \nonumber \\&\quad + c \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert \nonumber \\\le & {} c \Vert \, ^t Pu \Vert _{H^{0,s}} + \varepsilon c \left( \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \right) . \end{aligned}$$
(48)

By using (34), (39), (48) and Lemma 4, it results

$$\begin{aligned} \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \le c \Vert \, ^t Pu \Vert _{H^{0,s}} + c \varepsilon \Vert u \Vert _{H^{0,s}}. \end{aligned}$$

For \(\varepsilon \) small enough and making use of Lemma 4, we have

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} \\&\quad \le c (\Vert \, ^t P u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}}) \\&\quad \le c (\Vert \, ^t P u \Vert _{H^{0,s}} + \Vert (x_0 - \overline{x}_0) \partial _{x_0} u \Vert _{H^{0,s}}). \end{aligned}$$

For \(|x_0 - \overline{x}_0|\) small enough and using Lemma 4, we deduce

$$\begin{aligned}&\quad \ \, \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert \, ^t P u \Vert _{H^{0,s}}, \qquad \qquad \nonumber \\&\qquad \forall u \in C^{\infty }_0 (\overline{\Omega }): \ \mathrm{supp} \, u \subseteq ]\overline{x}_0, \overline{x}_0+k[ \times \Omega _0. \end{aligned}$$
(49)

Since the function \(\varphi \) is the same for every functions u, then c does not depend on u but depends on the distance between \(\partial \widetilde{\Omega }'_0\) and \(\partial \widetilde{\Omega }_0\) and k is small enough.

Now, if \(x_0 =0\), for every \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq [0,k[ \times \Omega _{0}\), we set \(v_s= \varphi (x') A_s u\). Making use of Theorem 10, it results

$$\begin{aligned} \Vert \partial _{x_0} v_s \Vert + \Vert \partial _{x_1} v_s \Vert + \Vert (x_0-\alpha (x')) \partial _{x_2} v_s \Vert + \Vert v_s \Vert \le \varepsilon \Vert \, ^t P v_s \Vert . \end{aligned}$$
(50)

Proceeding as done above, we obtain the analogous inequality of (39):

$$\begin{aligned}&\Vert \partial _{x_0} v_s \Vert + \Vert \partial _{x_1} v_s \Vert + \Vert (x_0-\alpha (x')) \partial _{x_2} v_s \Vert \nonumber \\&\quad \ge \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} - c \Vert x_0 \, ^t P u \Vert _{H^{0,s}} \nonumber \\&\qquad - c \Vert x_0 \partial _{x_0} u \Vert _{H^{0,s}} - c \Vert x_0 u \Vert _{H^{0,s}}, \end{aligned}$$
(51)

where we used Lemma 1 instead of Lemma 4. Considering \(\Vert \, ^t P v_s \Vert \) and proceeding again as done before and taking into account Theorem 10 and Lemma 1, we have

$$\begin{aligned} \Vert \, ^t Pv_s \Vert \le c \left( \Vert \, ^t Pu \Vert _{H^{0,s}} + \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x') \partial _{x_2} u \Vert _{H^{0,s}} \right) . \end{aligned}$$
(52)

Moreover, using (50), (51) and (52), we obtain

$$\begin{aligned}&\quad \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0-\alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert P u \Vert _{H^{0,s}}, \qquad \quad \nonumber \\&\qquad \forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq [0, k[ \times \Omega _0. \end{aligned}$$
(53)

\(\square \)

8 Global estimates

In this section, we obtain fundamental global estimates in order to prove the existence of a solution to the Cauchy–Dirichlet problem (2).

Theorem 12

Let (i), (ii) and (iii) be satisfied. Then, for every \(k>0\) and \(s<0\) there exists \(c>0\) such that

$$\begin{aligned}&\quad \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}}+ \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert \, ^t Pu \Vert _{H^{0,s}}, \quad \nonumber \\&\qquad \ \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _k= [0,k[ \times \Omega _0. \end{aligned}$$
(54)

Moreover, for \(s=0\) and for every \(k>0\) there exists \(c>0\) such that

$$\begin{aligned}&\Vert \partial _{x_0} u \Vert + \Vert \partial _{x_1} u \Vert + \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert + \Vert u \Vert \le c \Vert \, ^t Pu \Vert ,\nonumber \\&\quad \ \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \overline{\Omega }_k= [0,k[ \times \overline{\Omega }_0, \ u|_S=0. \end{aligned}$$
(55)

Finally, for every \(k>0\) and \(s<0\) there exists \(c>0\) such that

$$\begin{aligned}&\ \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}}+ \Vert (x_0 - \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert \, ^t P[u] \Vert _{H^{0,s}}, \quad \nonumber \\&\ \qquad \forall u \in C_{0}^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \overline{\Omega }_k= [0,k[ \times \overline{\Omega }_0, \ u|_S=0, \end{aligned}$$
(56)

where \([u]= {\left\{ \begin{array}{ll} \begin{array}{ll} u, &{} \quad \mathrm{in} \ \Omega _k= [0,k[ \times \Omega _0 \\ 0, &{} \quad \mathrm{in} \ \Omega _k= [0,k[ \times (\mathbb {R}^2 \setminus \Omega _0) \end{array} \end{array}\right. }\).

Proof

Let \(k > 0\), let us set \(\Omega _k = [0, k[ \times \Omega _0\). For the compactness of \([0, k] \times \overline{\Omega }_0\), there exists a finite number of subsets \(\{ \Omega _1, \Omega _2, \ldots , \Omega _p \}\) of \(\Omega _k\), given by

$$\begin{aligned} \Omega _1 =[0,h_1[ \times \Omega _0, \ \Omega _2 =[h_1',h_2[ \times \Omega _0, \ldots , \ \Omega _p =[h_{p-1}',h_p[ \times \Omega _0, \end{aligned}$$

with \(h_0 =0\), \(h_p=h\), \(h_{i-1}< h_i'< h_i\), for every \(i=1,\ldots ,p\), and such that (33) holds in every \(\Omega _i\), for \(i=1,\ldots ,p\).

Let \(u \in C_0^{\infty }(\Omega _k)\), let \(\varphi \in C_0^{\infty }([0,h_1[)\), with \(\varphi \equiv 1\) on \([0,h_1'[\) and \(0 \le \varphi \le 1\) in \([0, h_1[\). Rewriting (33) for \(\varphi u\), it results

$$\begin{aligned}&\Vert \partial _{x_0} \varphi u \Vert _{H^{0,s}} + \Vert \partial _{x_1} \varphi u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} \varphi u \Vert _{H^{0,s}} + \Vert \varphi u \Vert _{H^{0,s}} \\&\quad \le c \Vert P \varphi u \Vert _{H^{0,s}} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert [P, \varphi ] u \Vert _{H^{0,s}} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} \varphi \partial _{x_0} u \Vert _{H^{0,s}} + c \Vert (\partial _{x_0}^2 \varphi ) u \Vert _{H^{0,s}} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} u \Vert _{H^{0,s}([h_1',h_1[ \times \Omega _0)} + c \Vert u \Vert _{H^{0,s}([h_1',h_1[ \times \Omega _0)} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} u \Vert _{H^{0,s}([h_1',h_2'[ \times \Omega _0)} + c \Vert u \Vert _{H^{0,s}([h_1',h_2'[ \times \Omega _0)} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} \varphi _1 u \Vert _{H^{0,s}([h_1',h_2[ \times \Omega _0)} + c \Vert \varphi _1 u \Vert _{H^{0,s}([h_1',h_2[ \times \Omega _0)}, \end{aligned}$$

where \(\varphi _1 \in C_0^{\infty }(\Omega _0)\) such that \(\mathrm{supp} \, \varphi _1 \subseteq [h_1',h_2[\), \(\varphi _1 \equiv 1\) in \([h_1',h_2'] \times \Omega _0\).

We can deduce that

$$\begin{aligned}&\Vert \partial _{x_0} \varphi _{i-1} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} \varphi _{i-1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} \varphi _{i-1} u \Vert _{H^{0,s}} + \Vert \varphi _{i-1} u \Vert _{H^{0,s}} \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} \varphi _i u \Vert _{H^{0,s}([h_i',h_{i+1}[ \times \Omega _0)} + c \Vert \varphi _i u \Vert _{H^{0,s}([h_i',h_{i+1}[ \times \Omega _0)}, \end{aligned}$$

where \(\varphi _0 = \varphi \) and \(\varphi _i \in C_0^{\infty } ([0,k[)\) such that \(\mathrm{supp} \, \varphi _i \subseteq [h_i',h_{i+1}[\), for every \(i = 1, \ldots , p\).

On the other hand, we have

$$\begin{aligned}&\Vert \partial _{x_0} \varphi _{p-1} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} \varphi _{p-1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} \varphi _{p-1} u \Vert _{H^{0,s}} + \Vert \varphi _{p-1} u \Vert _{H^{0,s}} \nonumber \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \Vert \partial _{x_0} \varphi _p u \Vert _{H^{0,s}(\Omega _p)} + c \Vert \varphi _p u \Vert _{H^{0,s}(\Omega _p)} \nonumber \\&\quad \le c \Vert P u \Vert _{H^{0,s}} + c \left( \Vert \partial _{x_0} u \Vert _{H^{0,s}(\Omega _p)} + \Vert u \Vert _{H^{0,s}(\Omega _p)} \right) \nonumber \\&\quad \le c \Vert P u \Vert _{H^{0,s}}. \end{aligned}$$
(57)

Using (33), (57) and proceeding by recurrence on i, we easily obtain

$$\begin{aligned} \Vert \partial _{x_0} \varphi _{i} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} \varphi _{i} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} \varphi _{i} u \Vert _{H^{0,s}} + \Vert \varphi _{i} u \Vert _{H^{0,s}} \le c \Vert P u \Vert _{H^{0,s}}, \end{aligned}$$

for \(i = 1, \ldots , p\). Taking into account the previous inequality, we have

$$\begin{aligned}&\quad \Vert \partial _{x_0} u \Vert _{H^{0,s}} + \Vert \partial _{x_1} u \Vert _{H^{0,s}} + \Vert (x_0- \alpha (x')) \partial _{x_2} u \Vert _{H^{0,s}} + \Vert u \Vert _{H^{0,s}} \le c \Vert P u \Vert _{H^{0,s}}, \nonumber \\&\qquad \forall u \in C_0^{\infty }(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq \Omega _k. \end{aligned}$$
(58)

For the arbitrariness of k, (58) holds for every \(u \in C_0^{\infty }(\overline{\Omega })\). The proof of (54) is thereby completed.

Furthermore, taking into account (31), we obtain (55).

Finally, we prove (56). Let \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq [0, k[ \times \overline{\Omega }_0\) and \(u|_S=0\). Let \(\{ u_n \}\) be a sequence in \(C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u_n \subseteq [0, k[ \times \overline{\Omega }_0\) and \(u_n \rightarrow u\) in \(H^{2,1}\). We have that \(u_n \rightarrow u\) and \(Pu_n \rightarrow P[u]\) in \(H^{0,s}\), for every \(s<0\). Hence, rewriting (54) for \(u_n\), for every \(n \in \mathbb {N}\), and passing to the limit as \(n \rightarrow + \infty \), we obtain (56). \(\square \)

9 Proof of Theorem 1

Let V be the subspace of \(L^2(\Omega _k)\), where \(\Omega _k= ]0,k[ \times \Omega _0\), made up of functions \(\psi = \, ^tPu\), with \(u \in C^{\infty }_0(\overline{\Omega })\) such that \(\mathrm{supp} \, u \subseteq [0,k[ \times \overline{\Omega }_0\) and \(u|_S=0\). Let us consider the functional

$$\begin{aligned} T(\psi ) = T( \, ^t Pu) =(f, u). \end{aligned}$$

It results

$$\begin{aligned} |T(\psi )|= & {} |T(\, ^t Pu)| \\= & {} |(f,u)| \\\le & {} \Vert f \Vert _{L^2(\Omega _k)} \Vert u \Vert _{L^2(\Omega _k)}. \end{aligned}$$

Making use of (55), we have

$$\begin{aligned} |T(\psi )|\le & {} c \Vert f \Vert _{L^2(\Omega _k)} \Vert \, ^t Pu \Vert _{L^2(\Omega _k)} \\= & {} c \Vert f \Vert _{L^2(\Omega _k)} \Vert \psi \Vert _{L^2(\Omega _k)} \\= & {} c' \Vert \psi \Vert _{L^2(\Omega _k)}, \end{aligned}$$

where \(c'=c \Vert f \Vert _{L^2(\Omega _k)}\). Therefore, it is possible to extend T as a linear continuous functional into \(L^{2}(\Omega _k)\). Making use of a representation theorem, there exists \(w \in L^2(\Omega _k)\) such that

$$\begin{aligned} T(v)=( w,v ), \quad \forall v \in L^2(\Omega _k). \end{aligned}$$

In particular, we have

$$\begin{aligned} T(\psi ) = T(\, ^tPu) = (w, \, ^t Pu) =(f,u), \quad \forall u \in C^{\infty }_0(\overline{\Omega }_k): \ u|_S=0. \end{aligned}$$

Hence, w is a solution in the sense of distributions to the equation

$$\begin{aligned} Pu=f, \quad \mathrm{in} \ \Omega _k. \end{aligned}$$

For the arbitrariness of k and since \(f \in L^2_{loc}(\overline{\Omega })\), Theorem 1 is proved.

10 Proof of Theorem 2

Let us denote by W the subspace of \(\mathcal {D}'([0,k[ \times \Omega _0)\) containing extensions of linear continuous functionals to functions \(\varphi \in C^{\infty }_0(\overline{\Omega }_k)\) such that \(\varphi |_S=0\), where \(\overline{\Omega }_k= [0,k[ \times \overline{\Omega }_0\). It results that \(P[u] \in W\), where \(u \in C^{\infty }_0(\overline{\Omega }_k)\) such that \(u|_S=0\) and \(u=0\) in \([0,k[ \times (\mathbb {R}^2 \setminus \overline{\Omega }_0)\). Moreover, we have

$$\begin{aligned} \langle \varphi , \, ^t P[u] \rangle = ( \varphi , \, ^t P [u]) = (\varphi , \, ^t Pu), \quad \forall \varphi \in C^{\infty }_0(\overline{\Omega }_k): \ \varphi |_S=0. \end{aligned}$$

Therefore, the distributions \(^tP[u]\) and \(^t Pu\) are equal in W. Let T be the functional defined into the subspace of W containing the distributions \(\psi = \, ^t P[u]\), for every \(u \in C^{\infty }([0,k[ \times \overline{\Omega }_0)\) such that \(u|_S=0\), given by

$$\begin{aligned} T(\psi ) = T( \, ^t P[u]) = (f,u). \end{aligned}$$

Making use of (56), it follows

$$\begin{aligned} \Vert T(\psi ) \Vert= & {} |T( \, ^t P[u] )| \\= & {} |(f,u)| \\\le & {} \Vert f \Vert _{H^{0,s}} \Vert u \Vert _{H^{0,-s}} \\\le & {} c \Vert \, ^t P [u] \Vert _{H^{0,-s}}, \quad \forall u \in C^{\infty }_0(\overline{\Omega }_k): \ u|_S=0. \end{aligned}$$

with \(s \le r\). Then, T can be extended in the subspace \(W'\) of W containing the distributions of W with finite \(H^{0,-s}(\overline{\Omega }_k)\)-norm. As a consequence, there exists \(w \in W'^*\), where \(W'^*\) is the topological dual of \(W'\), such that

$$\begin{aligned} T(\psi ) = T(\, ^t P[u]) = (w, \, ^tP [u]) =(f,u). \end{aligned}$$
(59)

On the other hand, it results \(w \in H^{0,s}(\Omega _k)\) and since

$$\begin{aligned} (\varphi , \, ^tP[u]) = (\varphi , \, ^t Pu), \quad \forall \varphi , u \in C^{\infty }_0([0,k[ \times \overline{\Omega }_0): \ \varphi |_S=0, \ u|_S=0, \end{aligned}$$

it follows for every \(\{ \varphi _n \} \subseteq C^{\infty }_0([0,+ \infty [ \times \overline{\Omega }_0)\) such that \(\varphi _n|_S=0\), \(\forall n \in \mathbb {N}\), and \(\varphi _n \rightharpoonup w\) in \(W'^*\),

$$\begin{aligned} (Pw, u)= & {} (w, \, ^tP [u]) \nonumber \\= & {} \lim _{n \rightarrow + \infty } (\varphi _n, \, ^tP[u]) \nonumber \\= & {} \lim _{n \rightarrow + \infty } (\varphi _n, \, ^tPu) \nonumber \\= & {} (w, \, ^tPu), \end{aligned}$$
(60)

we deduce that \(w|_S=0\) (see also below).

Taking into account (61) and (60), we get

$$\begin{aligned} (w, \, ^tPu) = (Pw, u) = (f,u), \quad \forall u \in C^{\infty }_0(\overline{\Omega }): \ \mathrm{supp} \, u \subseteq [0,k[ \times \overline{\Omega }_0. \end{aligned}$$
(61)

From (61), we have

$$\begin{aligned} Pw=f, \quad \mathrm{in \ the \ sense \ of \ distributions}. \end{aligned}$$

and

$$\begin{aligned} w \in H^r(\Omega _k \setminus \partial \Omega _0). \end{aligned}$$

Indeed, set \(Lw= Pw+ \partial _{x_0}^2 w - \frac{1}{i} a_0(x) \partial _{x_0} w - b(x) w\), it results

$$\begin{aligned} - \partial _{x_0}^2 w + \frac{1}{i} a_0(x) \partial _{x_0} w + b(x) w= f- Lw, \end{aligned}$$
(62)

with \(w \in \mathcal {D}'(\Omega _k) \cap H^{0,r}(\Omega _k)\) and \(f-Lw \in L^2(\Omega _k)\). From (62), it follows that w is a solution to a second-order differential equation with zero-order term belonging to \(L^2(\Omega _k)\). Hence, we have \(w \in H^{2,0}(\Omega _k) \cap H^{0,r}(\Omega _k)\). On the other hand, (62) implies

$$\begin{aligned} \partial ^{0, \alpha _1, \alpha _2} \left( - \partial _{x_0}^2 w + \frac{1}{i} a_0(x) \partial _{x_0} w + b(x) w \right) =\partial ^{0, \alpha _1, \alpha _2} \left( f- Lw \right) , \end{aligned}$$

with \(\alpha _1+\alpha _2 \le s-r+2\). Therefore, we obtain

$$\begin{aligned}&- \partial _{x_0}^2 \partial ^{0, \alpha _1, \alpha _2} w + \frac{1}{i} a_0(x) \partial _{x_0} \partial ^{0, \alpha _1, \alpha _2} w + b(x) \partial ^{0, \alpha _1, \alpha _2} w \nonumber \\&\quad = \partial ^{0, \alpha _1, \alpha _2} \left( f- Lw \right) + \left[ \partial ^{0, \alpha _1, \alpha _2} - \partial _{x_0}^2 + \frac{1}{i} a_0(x) \partial _{x_0} + b(x) \right] w. \end{aligned}$$
(63)

Proceeding by induction in the previous equality, assuming \(u \in H^{2, p-1}\), with \(1 \le p \le r-2\) and taking into account (63), it results

$$\begin{aligned} w \in H^{2,p}(\Omega _k \setminus \partial \Omega _0). \end{aligned}$$

Subsequently, by the equality

$$\begin{aligned} \partial ^{p-2, \alpha _1, \alpha _2} \left( - \partial _{x_0}^2 w + \frac{1}{i} a_0(x) \partial _{x_0} w + b(x) w \right) =\partial ^{p-2, \alpha _1, \alpha _2} \left( f- Lw \right) , \end{aligned}$$

with \(0 \le p-2+ \alpha _1 + \alpha _2 \le r-2\), and proceeding by induction on p, it follows

$$\begin{aligned} w \in H^{r}(\Omega _k\setminus \partial \Omega _0). \end{aligned}$$

From (61), we deduce

$$\begin{aligned} \langle Pw, u \rangle= & {} (Pw, u) \\= & {} ( w, \, ^t Pu ) \\= & {} (f, u), \quad \forall u \in C^{\infty }_0(\Omega _k): \ \mathrm{supp} \, u \subseteq ]0,k[ \times \Omega _0. \end{aligned}$$

Then, we obtain

$$\begin{aligned} Pw=f, \quad \mathrm{a.e. \ in} \ \mathrm{int} \, \Omega _k. \end{aligned}$$

Now, making use of (61), we show that the boundary conditions on \(\Omega _0\) are satisfied. Let \(u(x_0,x')=u_0(x_0)u_1(x')\) such that \(u_0 \in C^{\infty }_0([0,k_1[)\), \(u_0(0)=1\), \(\partial _{x_0} u_0(0)=0\) and \(u_1 \in C^{\infty }_0(\Omega _0)\). Integrating by parts in (61), we have

$$\begin{aligned} (Pw, u) - \int _{\Omega _0} w(0,x') u_1(x') {\text {d}}x' =(w, \, ^t Pu). \end{aligned}$$

It follows

$$\begin{aligned} \int _{\Omega _0} w(0,x') u_1(x') {\text {d}}x' =0, \quad \forall u_1 \in C^{\infty }_0(\Omega _0). \end{aligned}$$

It implies

$$\begin{aligned} w(0,x') =0, \quad \mathrm{a.e. \ in} \ \Omega _0. \end{aligned}$$

Instead, if \(u(x_0,x')= u_0(x_0)u_1(x')\), with \(u_0 \in C^{\infty }_0([0,k_1[)\), \(u_0(0)=0\), \(\partial _{x_0} u_0(0)=1\) and \(u_1 \in C^{\infty }_0(\Omega _0)\), integrating by parts, we obtain

$$\begin{aligned} \int _{\Omega _0} \partial _{x_0} w(0,x') u_1(x') {\text {d}}x' =0, \quad \forall u_1 \in C^{\infty }_0(\Omega _0). \end{aligned}$$

Hence, it results

$$\begin{aligned} \partial _{x_0} w(0,x')=0, \quad \mathrm{a.e. \ in} \ \Omega _0. \end{aligned}$$

Then, we have proved that the Cauchy problem

$$\begin{aligned} {\left\{ \begin{array}{ll} Pw=f, \quad \mathrm{in} \ \Omega _k, \\ w|_{\Omega _0}=0, \ \frac{{\text {d}}w}{{\text {d}}n}|_{\Omega _0}=0, \end{array}\right. } \end{aligned}$$

admits a solution \(w \in H^r(\overline{\Omega }_k \setminus \partial \Omega _0)\), for every \(k>0\), under assumptions (i), (ii) and (iii) and if \(f \in H^r(\overline{\Omega }_k)\). Finally, we justify that \(w|_S=0\), as written above. In fact, integrating by parts in (61), we get

$$\begin{aligned} (Pw,u) + \int _{S} w n_1 \partial _{x_1} u {\text {d}} \sigma + \int _{S} w n_2 (x_0-\alpha (x'))^2 \partial _{x_2} u {\text {d}} \sigma = (w, \, ^tPu). \end{aligned}$$

It follows

$$\begin{aligned} \int _{S} w (n_1 \partial _{x_1} u + n_2 (x_0-\alpha (x'))^2 \partial _{x_2} u ) {\text {d}} \sigma = 0. \end{aligned}$$

Fixed an arbitrary test function \(\phi \) on S, it is possible to determine u such that \(n_1 \partial _{x_1} u + n_2 (x_0- \alpha (x'))^2 \partial _{x_2} u|_S = \phi (x_0,x')\). Then, we obtain

$$\begin{aligned} \int _S w \phi {\text {d}} \sigma =0, \quad \forall \phi \in C^{\infty }_0(S), \end{aligned}$$
(64)

which implies

$$\begin{aligned} w=0, \quad \mathrm{a.e. \ in} \ S. \end{aligned}$$

In the following, a brief proof of the previous claim is given. Parameterizing the surface S in the following way:

$$\begin{aligned} x_0=x_0, \quad x_1= \varphi _1(s), \quad x_2= \varphi _2(s), \end{aligned}$$

with \(x_0 \in [0,k[\) and \(s \in [0, L(\partial \Omega _0)]\), being s the arc length of \(\partial \Omega _0\), we have

$$\begin{aligned}&\int _S w( n_1 \partial _{x_1}u + n_2 (x_0-\alpha (x'))^2 \partial _{x_2} u) {\text {d}} \sigma \\&\quad = \int _{[0,k] \times [0, L(\partial \Omega _0)]} w(x_0, \varphi _1(s), \varphi _2(s)) \varphi _2'(s) \partial _{x_1} u(x_0, \varphi _1(s), \varphi _2(s)) {\text {d}}x_0 {\text {d}}s \\&\quad - \int _{[0,k] \times [0, L(\partial \Omega _0)]} w(x_0, \varphi _1(s), \varphi _2(s)) \varphi _1'(s) (x_0-\alpha (\varphi _1(s),\varphi _2(s)))^2 \\&\quad \cdot \partial _{x_2} u(x_0, \varphi _1(s), \varphi _2(s)) {\text {d}}x_0 {\text {d}}s \\&\quad = \int _{[0,k] \times [0, L(\partial \Omega _0)]} w(x_0,s) \frac{du}{dn}(x_0,s) ((\varphi _2'(s))^2 + (x_0 - \alpha (s))^2 (\varphi _1'(s))^2) {\text {d}}x_0 {\text {d}}s, \end{aligned}$$

where n is the external normal vector to the surface S. Hence, in order to obtain (64), we need that \(\dfrac{du}{dn}|_S = \phi (x_0,s)\), where \(\phi \) is an arbitrary function belonging to \(C^{\infty }_0( [0,k[ \times ]0, L(\partial \Omega _0) [)\). As a consequence, we have proved the existence of a solution \(w \in H^r(\Omega _k \setminus \partial \Omega _0)\) to the following Cauchy–Dirichlet problem

$$\begin{aligned} {\left\{ \begin{array}{ll} Pw=f, \quad \mathrm{in} \ \Omega _k, \\ w|_{\Omega _0}=0, \ \frac{dw}{dn}|_{\Omega _0}=0, \ w|_S=0, \end{array}\right. } \end{aligned}$$

where \(f \in H^r(\overline{\Omega }_k)\). Since \(f \in H^{r}_{loc}(\overline{\Omega })\) and for the arbitrariness of k, Theorem 2 is obtained.