Boundary behavior and monotonicity estimates for solutions to nonlinear diffusion equations (Q1076905)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Boundary behavior and monotonicity estimates for solutions to nonlinear diffusion equations
scientific article

    Statements

    Boundary behavior and monotonicity estimates for solutions to nonlinear diffusion equations (English)
    0 references
    1985
    0 references
    The paper deals with the parabolic equation \[ \nu_ t=A(\nu)\nu_{xx}\quad on\quad (0,1)\times {\mathbb{R}}^+,\quad \nu =0\quad on\quad (0,1)\times \{0\}, \] supplemented with boundary conditions on \(\{0\}\times {\mathbb{R}}^+\) and on \(\{1\}\times {\mathbb{R}}^+\) either of Dirichlet (temperature controlled case) or of Neumann type (flux controlled case). This problem is substantially equivalent to the more standard equation in divergence form \(u_ t=(a(u)u_ x)_ x\). After establishing certain compatibility conditions between Dirichlet and Neumann data, the author proves i) a-priori estimates for the solution and its space gradient; ii) monotone dependence of the boundary values on the coefficient A. Specifically, if \[ \partial_ t\nu_ i=A_ i(\nu)\partial_{xx}\nu_ i(x,t),\quad 0<x<1,\quad 0<t<T,\quad \nu_ i=0\quad on\quad (0,1)\times \{0\}, \] \[ -\partial_ x\nu_ i(0,t)=g_ 0(t);\quad \partial_ x\nu_ i(1,t)=0,\quad 0<t<T \] for \(i=1,2\), and the coefficients \(A_ i\) satisfy, beside certain general properties, \(A_ 1(0)=A_ 2(0)\), \(A_ 1'(0)>A_ 2'(0)\), then for \(0<\sigma_ 1<\sigma_ 2\) there exist \(t_ 1=t_ 1(\sigma_ 1)\), \(T_ 1=T_ 1(\sigma_ 2)\) such that \(0<t_ 1<T_ 1\) and \[ (a)\quad \nu_ 1(0,t)-\nu_ 2(0,T)\geq C(t-t_ 1)\lambda \quad for\quad t_ 1\leq t\leq T_ 1, \] where \(C>0\) and \(\lambda =\min_{s\in [\sigma_ 1,\sigma_ 2]}(A_ 1(s)-A_ 2(s));\) moreover, (b) for \(\sigma >0\) there exists \(\tau =\tau (\sigma)>0\) such that \[ \nu_ 1(0,t)-\nu_ 2(0,t)\leq K(\tau)\mu \quad for\quad 0\leq t\leq \sigma, \] where \(K(\tau)>0\), and \(\mu =\max_{s\in [0,\sigma]}(A_ 1(s)-A_ 2(s))\). a) shows that identical flux controlled experiments involving distinct coefficients \(A_ 1\), \(A_ 2\) in the appropriate class cannot produce identical Dirichlet data at \(x=0\); b) then shows that identical flux controlled experiments with identical coefficients must produce identical Dirichlet boundary values at \(x=0\). Similar results prevail in the temperature controlled case. These properties, which are proved in a careful and detailed way, are relevant when analysing inverse problems for nonlinear parabolic equations.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    equation in divergence form
    0 references
    compatibility conditions
    0 references
    a-priori estimates
    0 references
    monotone dependence
    0 references
    inverse problems
    0 references
    nonlinear parabolic equations
    0 references
    0 references
    0 references