Smoothness of solutions to the initial-boundary value problem for the telegraph equation on the half-line with a locally summable potential

S. A. Simonov · 2025

Read Smoothness of solutions to the initial-boundary value problem for the telegraph equation on the half-line with a locally summable potential by S. A. Simonov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-09397.txt.

Introduction

Sergey Simonov studies how smooth the waves are for a wire-like system on a half line. The work dates to 2025. The system starts at rest and is driven from its end point. A weight term along the line can be rough, with just a mild sum trait.

Simonov proves the wave at any set time lies in a space of smooth maps. He also shows the map from the drive signal to the wave is a two-way match in one space. When the weight term is a bit smoother, the match holds in a smoother space too. The result firms up control theory for such wave systems.

Catalog record

CREATORS. A. Simonov
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-09397

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Smoothness of solutions to the initial-boundary value problem for the telegraph equation on the half-line with a locally summable potential S. A. Simonov Abstract We study solutions to the system u t ​ t − u x ​ x + q ⁡ ( x ) ​ u = 0 , \displaystyle u_{tt}-u_{xx}+q(x)u=0, x > 0 , t > 0 , \displaystyle x>0,\ t>0, u | t = 0 = u t | t = 0 = 0 , \displaystyle u|_{t=0}=u_{t}|_{t=0}=0, x ⩾ 0 , \displaystyle x\geqslant 0, u | x = 0 = g ⁡ ( t ) , \displaystyle u|_{x=0}=g(t), t ⩾ 0 , \displaystyle t\geqslant 0, with a locally summable Hermitian matrix-valued potential q q and a 𝒞 ∞ \mathcal{C}^{\infty} -smooth ℂ n \mathbb{C}^{n} -valued boundary control g g vanishing near the origin. We prove that the solution u g ​ ( ⋅ , T ) u^{g}(\cdot,T) is a function from 𝒲 1 2 ​ ( [ 0 , T ] , ℂ n ) \mathcal{W}^{2}_{1}([0,T];\mathbb{C}^{n}) and that the control operator W T : g ↦ u g ​ ( ⋅ , T ) W^{T}:g\mapsto u^{g}(\cdot,T) is an isomorphism in ℒ 2 ​ ( [ 0 , T ] , ℂ n ) \mathcal{L}_{2}([0,T];\mathbb{C}^{n}) , and, in the case q ∈ ℒ 2 ​ ( [ 0 , T ] , ℂ n ) q\in\mathcal{L}_{2}([0,T];\mathbb{C}^{n}) , also an isomorphism in ℋ 2 ​ ( [ 0 , T ] , ℂ n ) \mathcal{H}^{2}([0,T];\mathbb{C}^{n}) . 1 Introduction Let L_{0} be a symmetric positive definite completely non-self-adjoint operator in a Hilbert space \mathscr{H} . Consider a dynamical system with boundary control, \alpha , \displaystyle u(t)\in{\rm Dom\,}L_{0}^{*}, \displaystyle t\geqslant 0, (1) \displaystyle u^{\prime\prime}(t)+L_{0}^{*}u(t)=0, \displaystyle t>0, (2) \displaystyle u(0)=u^{\prime}(0)=0, (3) \displaystyle\Gamma_{1}u(t)=f(t), \displaystyle t\geqslant 0. (4) Here \Gamma_{1} is the first of the two boundary operators [ 11 ] of the so-called Vishik’s boundary triple (\mathscr{K};\Gamma_{1},\Gamma_{2}) , \mathscr{K}:={\rm Ker\,}L_{0}^{*} , \dim\mathscr{K}=n_{\pm}(L_{0}) (the defect indices), \Gamma_{1},\Gamma_{2}:{\rm Dom\,}L_{0}^{*}\to\mathscr{K} , \Gamma_{1}:=L^{-1}L_{0}^{*}-I , \Gamma_{2}:=P_{\mathscr{K}}L_{0}^{*} ( P_{\mathscr{K}} is the projection on \mathscr{K} in \mathscr{H} and L is the Friedrichs extension of L_{0} ). The construction of this boundary triple is based on the Vishik’s decomposition [ 14 ] {\rm Dom\,}L_{0}^{*}={\rm Dom\,}L_{0}\dot{+}\mathscr{K}\dot{+}L^{-1}\mathscr{K} . The boundary control f is a function from \mathcal{C}^{\infty}([0,\infty);\mathscr{K}) with {\rm supp\,}f\subset(0,\infty) . In the boundary control approach to inverse problems (the BC-method, [ 2 , 3 , 4 , 5 ] ) one usually considers a family (a nest ) of subsystems \alpha^{T} of the system \alpha on finite intervals t\in[0,T] . For each T>0 one is in particular interested in the “smooth waves” which are solutions of \alpha at the time T , u^{f}(T)\in\mathscr{H} . Smooth waves form the linear manifold \dot{\mathscr{U}}^{T} which is called the reachable set , and its closure is called the reachable subspace \mathscr{U}^{T} of the system \alpha^{T} . Smooth controls form the class \dot{\mathscr{F}}^{T}:=\{f\in\mathcal{C}^{\infty}([0,T];\mathscr{K})\,|\,{\rm supp\,}f\subset(0,T]\} , which is dense in the outer space \mathscr{F}^{T}:=\mathcal{L}_{2}([0,T];\mathscr{K}) of \alpha^{T} . The space \mathscr{H} plays the role of the inner space of \alpha^{T} , and the correspondence between controls and solutions is established by the control operator W^{T}:\mathscr{F}^{T}\to\mathscr{H} acting by the rule f\mapsto u^{f}(T) on {\rm Dom\,}W^{T}=\dot{\mathscr{F}}^{T} . Properties of the control operator are determined by the operator L_{0} and play an important role in the BC-method. For example, for the problem of electromagnetic tomography W^{T} is unbounded [ 4 ] , for the problems of acoustic tomography W^{T} is bounded, but can be not invertible for T large enough [ 4 ] , and for the Schrödinger operator on the half-line W^{T} is an isomorphism from \mathscr{F}^{T} to \mathscr{U}^{T} (i. e., a bounded operator which has a bounded inverse from \mathscr{U}^{T} to \mathscr{F}^{T} ) [ 7 ] . The nest of projections on reachable subspaces \{P_{\mathscr{U}^{T}}\}_{T\in[0,\infty]} serves as a resolution of the identity for the eikonal operator E=\int_{[0,\infty)}TdP_{\mathscr{U}^{T}} in \mathscr{H} . Passing to the spectral representation of this operator, we obtain the model space of functions where the operator L_{0} has its own representation \tilde{L}_{0} , which is considered as a functional model of L_{0} [ 9 , 8 ] . From many possible spectral representations the one based on the triangular factorization [ 12 ] of the connecting operator C^{T}:=(W^{T})^{*}W^{T} with respect to the nest of delayed control subspaces \{\mathscr{F}^{T}_{s}\}_{s\in[0,T]} , \mathscr{F}^{T}_{s}:=\mathcal{L}_{2}([T-s,T];\mathscr{K}) , plays a distinguished role, because it leads to a local functional model of L_{0} , which provides a way to solve inverse problems. It is based on finding the diagonal D_{W^{T}} of the control operator [ 6 ] with respect to the nest \{\mathscr{F}^{T}_{s}\}_{s\in[0,T]} . This is why the properties of the control operator are very important in the BC-method. In the present paper we consider the minimal matrix Schrödinger operator L_{0}=-\frac{d^{2}}{dx^{2}}+q(x) in the Hilbert space \mathscr{H}=\mathcal{L}_{2}([0,\infty);\mathbb{C}^{n}) with a locally summable Hermitian matrix-valued potential q and study properties of the corresponding smooth waves and the control operator. For such L_{0} we show that the system \alpha^{T} takes a form of the system \alpha^{T}_{\rm v} : \displaystyle u(\cdot,t)\in{\rm Dom\,}L_{0}^{*}, \displaystyle t\in[0,T], (5) \displaystyle\ddot{u}(x,t)-u_{xx}(x,t)+q(x)u(x,t)=0, \displaystyle x\in(0,\infty),t\in(0,T), (6) \displaystyle u(x,0)=\dot{u}(x,0)=0, \displaystyle x\in[0,\infty), (7) \displaystyle u(0,t)=f_{\rm v}(t), \displaystyle t\in[0,T], (8) where f_{\rm v}\in\mathscr{F}^{T}_{\rm v}:=\mathcal{L}_{2}([0,T];\mathbb{C}^{n}) is simply related to the control f so that its solution u^{f_{\rm v}}_{\alpha}(x,t) coincides with u^{f}(x,t) (see ( 14 ) below) and the differentiation has to be understood in the following sense: \dot{u} as a derivative of an \mathscr{H} -valued function in t and u_{x} as a partial derivative of a vector-valued function of two variables x and t with respec to to x . The system \alpha^{T}_{\rm v} is almost the same as the initial-boundary value problem \beta^{T}_{\rm v} for the telegraph equation \displaystyle u_{tt}(x,t)-u_{xx}(x,t)+q(x)u(x,t)=0, \displaystyle x\in(0,\infty),t\in(0,T), (9) \displaystyle u(x,0)=u_{t}(x,0)=0, \displaystyle x\in[0,\infty), (10) \displaystyle u(0,t)=f_{\rm v}(t), \displaystyle t\in[0,T], (11) where u_{t} denotes a partial derivative with respect to t . If the potential q and the control f_{\rm v} are both smooth and the control vanishes near 0 , it is well known that the system ( 9 )–( 11 ) has the classical solution u^{f_{\rm v}}_{\beta}(x,t):=f_{\rm v}(t-x)+\int_{x}^{t}w(x,s)f_{\rm v}(t-s)ds, (12) where w is a smooth kernel. One can see that this solution also solves the system \alpha^{T}_{\rm v} and thus W^{T}f=u^{f}(T)=u^{f_{\rm v}}_{\alpha}(T)=u^{f_{\rm v}}_{\beta}(T) holds. This defines the corresponding control operator W^{T}_{\rm v}:f_{\rm v}\mapsto u^{f_{\rm v}}_{\alpha}(T) , which is an isomorphism in \mathcal{L}_{2}([0,T];\mathbb{C}^{n}) (this is clear from ( 12 )). Therefore W^{T} (its closure) is an isomorphism from \mathscr{F}^{T}=\mathcal{L}_{2}([0,T];\mathscr{K}) to \mathscr{U}^{T}=\mathcal{L}_{2}([0,T];\mathbb{C}^{n}) . For smooth contols f(t) the solution u^{f}(\cdot,T) is smooth, and if one considers the restiction of W^{T}_{\rm v} to \mathcal{H}^{2}([0,T];\mathbb{C}^{n}) , this operator turns out to be an isomorphism in this space (in the norm of this space). Note that in the general situation one has the abstract representation u^{f}(t)=-f(t)+\int_{0}^{t}L^{-\frac{1}{2}}\sin(L^{\frac{1}{2}}(t-s))f^{\prime\prime}(s)ds (13) for the solution of the system \alpha^{T} , but it does not help with establishing of the properties of the smooth waves discussed above. However, it helps to see that u^{f}(t) is unique and smooth in t as an \mathscr{H} -valued function. In the present work we consider potentials q from the class \mathcal{L}_{1,\rm loc}([0,\infty);\mathbb{M}^{n}_{\mathbb{C}}) , where \mathbb{M}^{n}_{\mathbb{C}} denotes square matrices of size n with complex entries. In Theorem 1 we show that, although the kernel w(x,t) is only \mathcal{W}^{1}_{1} -smooth in x ,

Read the complete text