Solution regularity and smooth dependence for abstract equations and applications to hyperbolic PDEs

TL;DR

This paper develops a generalized implicit function theorem for abstract equations with limited smoothness assumptions and applies it to hyperbolic PDEs, establishing solution regularity and smooth dependence on parameters.

Contribution

It introduces a new implicit function theorem for equations with partial smoothness and applies it to hyperbolic PDEs, addressing small divisor issues and smooth dependence.

Findings

Established conditions to prevent small divisors in hyperbolic PDEs.

Proved smooth dependence of solutions on parameters under certain conditions.

Applied the theory to time-periodic solutions of hyperbolic systems.

Abstract

In the first part we present a generalized implicit function theorem for abstract equations of the type $F(\lambda,u)=0$. We suppose that $u_0$ is a solution for $\lambda=0$ and that $F(\lambda,\cdot)$ is smooth for all $\lambda$, but, mainly, we do not suppose that $F(\cdot,u)$ is smooth for all $u$. Even so, we state conditions such that for all $\lambda \approx 0$ there exists exactly one solution $u \approx u_0$, that $u$ is smooth in a certain abstract sense, and that the data-to-solution map $\lambda \mapsto u$ is smooth. In the second part we apply the results of the first part to time-periodic solutions of first-order hyperbolic systems of the type $$ \partial_tu_j + a_j(x,\lambda)\partial_xu_j + b_j(t,x,\lambda,u) = 0, \; x\in(0,1), \;j=1,\dots,n $$ with reflection boundary conditions and of second-order hyperbolic equations of the type $$ \partial_t^2u-a(x,\lambda)^2\partial^2_xu+b(t,x,\lambda,u,\partial_tu,\partial_xu)=0, \; x\in(0,1) $$ with mixed boundary conditions (one Dirichlet and one Neumann). There are at least two distinguishing features of these results in comparison with the corresponding ones for parabolic PDEs: First, one has to prevent small divisors from coming up, and we present explicit sufficient conditions for that in terms of $u_0$ and of the data of the PDEs and of the boundary conditions. And second, in general smooth dependence of the coefficient functions $b_j$ and $b$ on $t$ is needed in order to get smooth dependence of the solution on $\lambda$, this is completely different to what is known for parabolic PDEs.