Evolutionary equations with state-dependent delay

Bernhard Aigner, Marcus Waurick · 2025

Read Evolutionary equations with state-dependent delay by Bernhard Aigner, Marcus Waurick (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-2511-14883.txt.

Introduction

Bernhard Aigner and Marcus Waurick extend a key existence proof to a broad class of evolving equations with delays that depend on the state itself. In such problems, how long the delay lasts is set by the current solution, which makes the math much harder. The work sits inside the evolutionary equations framework built around material laws and unbounded operators.

The authors prove local well-posedness for weak solutions, given smooth starting histories and a few consistency conditions. They test the result on classic cases such as heat, wave, and Maxwell equations, plus systems with memory effects. The proof gives a shared foundation for many delayed physical models.

Catalog record

CREATORBernhard Aigner, Marcus Waurick
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE176 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-14883

Browse the full library · About the Tower of Babel

Full text

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

Evolutionary equations with state-dependent delay Thanks: supported by the state of Saxony via a graduate student stipend Bernhard Aigner, Marcus Waurick Abstract . We extend a contraction mapping argument for ordinary state-dependent delay differential equations to evolutionary partial differential equations in the sense of R. Picard, that is, to equations of the form ( ∂ t M ( ∂ t ) + A ) u ( t ) = F ( t , u ( t ) ) \bigl(\partial_{t}M(\partial_{t})+A\bigr)u(t)=F\bigl(t,u_{(t)}\bigr) , where A A is an m \mathrm{m} -accretive (unbounded) linear operator and M M is a material law. We establish local well-posedness (in the sense of weak solutions) of generalized initial value problems that stem from a distributional formulation. We require prehistories in H 1 H^{1} with bounded derivative, a regularity increasing right-hand side and a consistency condition. We showcase the viability of our results by applying them to classical examples (heat, wave and Maxwell’s equations), examples from semigroup theory, port-Hamiltonian systems, as well as equations featuring fractional derivatives and convolutions (in time) with bounded operators. 1. Introduction In this article we study partial differential equations (PDEs) with state-dependent delay, which are PDEs of the form (Du)(t)=F\bigl(t,u_{(t)}\bigr)\qquad t>0\text{,} where D is a (partial) differential operator in space-time and the right-hand side F is a function depending on the history u_{(t)}\colon\,[-h,0]\rightarrow H\text{,}\qquad s\mapsto u(t+s) of the state u . The parameter h>0 describes a suitable backward time-horizon and H at this point denotes any Hilbert space. The dependence on prior histories demands — in contrast to equations without delay — a prescribed initial prehistory \Phi instead of traditional initial values: u_{(0)}=\Phi\qquad\Longleftrightarrow\qquad u(t)=\Phi(t)\quad\text{for }-h\leq t\leq 0\text{.} This article follows up on recent discoveries concerning the solution theory of initial value problems (IVPs) for ordinary differential equations (ODEs) (i.e., D=\nicefrac{{\operatorname{d}}}{{\operatorname{dt}}} ) with state-dependent delay utilizing a weak solution theory and exponentially weighted Sobolev spaces. We refer to [ 11 ] for the principal publication and to [ 1 ] for a summary of the main results and some elementary examples. We will extend the approach presented there to evolutionary equations in the sense of R. Picard, i.e., PDEs of the form (1) \bigl(\partial_{t}M(\partial_{t})+A\bigr)u(t)=f(t)+F\bigl(t,(u+Z)_{(t)}\bigr)\qquad\,t>0\text{,}\\ where M is a holomorphic operator-valued function that is bounded on a right-half space, M(\partial_{t}) is a so-called material law operator (cf. section 2.2 or [ 32 , ch. 5.2] ) and A is an \mathrm{m} -accretive operator on the Hilbert space H . In particular, our results hold for skew-selfadjoint A . The function f is assumed to be suitably regular and integrable, Z is an extension of some initial prehistory \Phi\in H^{1}(-h,0;H) and F\colon[0,\infty)\times L_{2}(-h,0;H)\to H is a continuous map incorporating the state-dependent delay u_{(t)} . In the theory of evolutionary equations, IVPs of the form (2) \displaystyle\begin{aligned} \bigl(\partial_{t}M(\partial_{t})+A\bigr)u(t)&=F\bigl(t,u_{(t)}\bigr)&&\qquad\qquad\,t>0\text{,}\\ u(t)&=\Phi(t)&&\quad-h\leq t\leq 0\text{,}\end{aligned} are equivalent to a reformulation as evolutionary equations on the entire line without initial value, albeit in a distributional setting, cf. [ 32 , thm. 9.4.3] . The term f in eq. 1 originates from this reformulation and suitably incorporates the initial prehistory \Phi . For more complex material laws, such a formulation is obtained using a projection argument. We provide some background in section 2.3 . The appearance of a distributional formulation is also present in other approaches. Indeed, in the context of solution theory utilizing semigroup theory, well-posedness is understood in the sense of existence of mild solutions. It is a well-known fact, that mild solutions of a Cauchy problem are the classical solutions of the extrapolated Cauchy problem (cf. [ 9 , sec. VI.7] or [ 31 , sec. 2.2] ). In the context of delay differential equations (DDEs), semigroup theory has been utilized extensively, e.g., cf. [ 2 ] for a comprehensive overview and also [ 44 ] . There are results for the non-autonomous setting (cf. [ 12 ] ) or with asymptotically autonomous delay (cf. [ 30 ] ). Here, we offer a different perspective, but we compare our results to semigroup approaches in section 6.4 and section 6.5 . In contrast to semigroup theory, the theory of evolutionary equations can make short work of mixed type problems (e.g., mixed eddy current models, cf. section 6.3 ) and the material law can incorporate operators that are even nonlocal in time (cf. section 6.7 ). The large applicability of evolutionary equations comes at the price of a more involved problem formulation for initial value problems. Indeed, partial differential algebraic equations are included; a subclass for which it is already a challenging task to identify a suitable space of initial values. We choose to forego this problem and focus on the well-posedness aspect. This approach is reminiscent of the foundational article of evolutionary equations [ 24 ] and the works [ 37 , 38 , 39 ] of S. Trostorff. We want to transform the problem ( 2 ) into a fixed point problem (FPP) and then utilize a contraction mapping argument. To that end, we employ weighted Lebesgue and Sobolev spaces: Definition 1.1 . For \rho>0 and -\infty\leq a<b\leq+\infty let \displaystyle L_{2,\rho}(a,b;H) \displaystyle\coloneq\overline{\mathcal{C}_{c}^{\infty}\bigl(a,b;H\bigr)}^{\lVert\cdot\rVert_{2,\rho}}\text{,} \displaystyle\lVert u\rVert_{2,\rho} \displaystyle\coloneq\Bigl(\mathop{\mathord{\raise 0.49991pt\hbox{$\displaystyle\int$}}}\nolimits_{\kern-4.00002pta}^{b}\lVert u(t)\rVert^{2}_{H}\mathrm{e}^{-2\rho t}\,\mathrm{d}{}t\Bigr)^{\nicefrac{{1}}{{2}}}\text{,} and, accordingly, \displaystyle H_{\rho}^{1}(a,b;H) \displaystyle\coloneq\bigl\{u\in L_{2,\rho}(a,b;H)\colon\,u^{\prime}\in L_{2,\rho}(a,b;H)\bigr\}\text{,} \displaystyle\lVert u\rVert_{H^{1}_{\rho}} \displaystyle\coloneq\bigl(\lVert u\rVert^{2}_{2,\rho}+\lVert u^{\prime}\rVert^{2}_{2,\rho}\bigr)^{\nicefrac{{1}}{{2}}}\text{,} where \mathcal{C}_{c}^{\infty}\bigl(a,b;H\bigr) is the space of all smooth functions with values in H and compact support in (a,b) . The derivative in the definition is to be understood in the distributional sense. The use of weighted Sobolev spaces, rather than the classical (unweighted) versions will allow us to force contractivity of a Lipschitz-mapping, reminiscent of a classical proof of the Picard–Lindelöf theorem for ODEs by Morgenstern, cf. [ 20 ] . For a general introduction to Sobolev spaces we refer to [ 10 , ch. 5] , for exponentially weighted Sobolev spaces to [ 32 , ch. 3] . The article is structured as follows: We begin with a short recap on evolutionary equations and mostly recall established results with a few new considerations. We follow up with some observations on the delay and on the nature of evolutionary equations with state-dependent delay. In section 4 we prove our main (well-posedness) results, which we make more explicit for the case of simple material laws in section 5 . We conclude with a large variety of examples in section 6 and a short review of our results paired with a comparison to other results, in particular to results utilizing semigroup theory. The appendix contains definitions of differential operators employed in section 6 and recapitulates known results for the formulation of IVPs in the framework of evolutionary equations. 2. Background on evolutionary equations In the context of evolutionary equations, the left-hand side of eq. 1 is split up into two parts, on one hand the “material dependence” \partial_{t}M(\partial_{t}) , which is time-dependent, and on the other hand a purely spatial part A , which is assumed to be a densely defined, linear, \mathrm{m} -accretive operator on the Hilbert space H . To explain the first part, we first need to explain a central concept in evolutionary equations, the time derivative in exponentially weighted spaces. 2.1. The time derivative Let \rho>0 and -\infty<a<b\leq+\infty . Let \displaystyle I_{\rho}\colon\,L_{2,\rho}(a,b;H) \displaystyle\to L_{2,\rho}(a,b;H)\text{,} \displaystyle u \displaystyle\mapsto\Bigl(t\mapsto\mathop{\mathord{\raise 0.49991pt\hbox{$\displaystyle\int$}}}\nolimits_{\kern-4.00002pta}^{t}u(s)\,\mathrm{d}{}s\Bigr)\text{.} This operator is well-defined on L_{2,\rho}(a,b;H) , as the next

Read the complete text