\(C^{\infty}\) ill-posedness of the mixed problem for hyperbolic equations of second order (Q1100347)

From MaRDI portal
scientific article
Language Label Description Also known as
English
\(C^{\infty}\) ill-posedness of the mixed problem for hyperbolic equations of second order
scientific article

    Statements

    \(C^{\infty}\) ill-posedness of the mixed problem for hyperbolic equations of second order (English)
    0 references
    0 references
    1988
    0 references
    We consider a mixed problem \[ (1)\quad L[u]=D^ 2_ tu-2\sum ^{n}_{j=1}h_ j(t,x)D_ tD_{x_ j}u-\sum ^{n}_{i,j=1}a_{ij}(t,x)D_{x_ i}D_{x_ j}u+(lower\quad order)=f(t,x)\quad in\quad \Omega \] \[ B[u]| _{x_ 1=0}=a_{11}(t,0,x')^{-1/2}\{\sum ^{n}_{j=1}a_{1j}(t,0,x')D_{x_ j}u+h_ 1(t,0,x')D_ tu\}+ \] \[ +\sum ^{n}_{j=1}b_ j(t,x')D_{x_ j}u-c(t,x')(1+h_ 1(t,0,x')^ 2/a_{11}(t,0,x'))^{1/2}\{D_ tu-(1+ \] \[ +h_ 1(t,0,x')^ 2/a_{11}(t,0,x'))^{-1}\sum ^{n}_{j=2}(h_ j(t,0,x')- \] \[ -h_ 1(t,0,x')/a_{11}(t,0,x')a_{1j}(t,0,x'))D_{x_ j}u\} \] \[ +\gamma (t,x')u| _{x_ 1=0}=g(t,x')\quad in\quad {\hat \Omega},\quad \sup p[u]\subset \Omega ^ +_{T_ 0} \] where \(x=(x_ 1,x_ 2,...,x_ n)\), \(x'=(x_ 2,...,x_ n)\) (n\(\geq 2)\), \(D_ t=\partial /\partial t\), \(D_{x_ j}=\partial /\partial x_ j\), \(\Omega\) be an open neighborhood of \((t,x_ 1,x')=(0,0,0,...,0)\) in \({\mathbb{R}}^ 1\times \overline{{\mathbb{R}}^ 1_ +}\times {\mathbb{R}}^{n-1}\), \({\hat \Omega}=\{(t,x')|\) (t,0,x')\(\in \Omega \}\) and \(\Omega ^ +_{T_ 0}=\{(t,x_ 1,x')\in \Omega |\) \(t\geq T_ 0\}\). We set (2) \[ b(t,x',\eta ')=\sum ^{n}_{j=2}b_ j(t,x')\eta _ j/\tilde d(t,x',\eta ') \] \[ \tilde d(t,x',\eta ')=[\sum ^{n}_{i,j=2}a_{ij}(t,0,x')\eta _ i\eta _ j-(\sum ^{n}_{j=2}a_{1j}(t,0,x')\eta _ j)^ 2/a_{11}(t,0,x')+ \] \[ +(1+h_ 1(t,0,x')^ 2/a_{11}(t,0,x'))^{-1}\{\sum ^{n}_{j=2}(h_ j(t,0,x')- \] \[ -(h_ 1(t,0,x')/a_{11}(t,0,x'))a_{1j}(t,0,x'))\eta _ j\}^ 2]^{1/2}. \] We assume the following condition for the problem (1): (A.I) L is regularly hyperbolic in \(\Omega ^ +_{T_ 0}\), \(a_{11}(t,0,x')>0\) in \({\hat \Omega}{}^ +_{T_ 0}=\{(t,x')|\) \((t,0,x')\in \Omega ^ +_{T_ 0}\}\) and \({\hat \Omega}{}^ +_{T_ 0}\) is connected. (A.II) c(t,x')\(\neq -1\) in \({\hat \Omega}{}^ +_{T_ 0}\). (A.III) The quadratic equation \[ (c(T,x_ 0')+1)Z^ 2+2b(T,x_ 0',\eta _ 0')Z+c(T,x_ 0')-1=0 \] has two roots \(Z_ 1\) and \(Z_ 2\) such that \(| Z_ 2| >1\) and Im \(Z_ 2\neq 0\) for \((T,x_ 0')\in {\hat \Omega}^ +_{T_ 0}.\) We obtain the following result: Assume the conditions (A.I), (A.II) and (A.III). Then, the mixed problem (1) is not \(C^{\infty}\)-well-posed in \(\Omega ^ +_{T_ 0}.\) We remark that the mixed problem (1) is \(L^ 2\)-well-posed in \(\Omega ={\mathbb{R}}^ 1\times \overline{{\mathbb{R}}^ 1_ +}\times {\mathbb{R}}^{n-1}\) if and only if for all (t,x')\(\in {\hat \Omega}={\mathbb{R}}^ 1\times {\mathbb{R}}^{n-1}\), the quadratic equation \[ (c(t,x')+1)Z^ 2+2b(t,x',\eta ')Z+c(t,x')-1=0 \] has roots in \=D\(=\{Z\in {\mathbb{C}}|\) \(| Z| \leq 1\}\) if they are different and in \(D=\{Z\in {\mathbb{C}}|\) \(| Z| <1\}\) if they are equal.
    0 references
    0 references
    0 references
    0 references
    0 references
    ill-posedness
    0 references
    mixed problem
    0 references
    0 references