Exact output tracking for the one-dimensional heat equation and applications to the interpolation problem in Gevrey classes of order 2
Lucas Davron, and Pierre Lissy · 2025
Read Exact output tracking for the one-dimensional heat equation and applications to the interpolation problem in Gevrey classes of order 2 by Lucas Davron, and Pierre Lissy (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-2511-07326.txt.
Introduction
Lucas Davron and Pierre Lissy ask which outputs a heated rod can be forced to follow exactly. They study the one-dimensional heat equation, steered by control at one boundary, and ask which temperature traces at the other boundary are reachable. The answer depends on smoothness over finite or endless time.
For endless time, reachable traces are exactly those whose derivatives meet a weighted summability rule tied to Gevrey classes of order two. Finite time adds a compatibility condition linked to reachable states. Fourier and Laplace tools, Hardy spaces, and a new Plancherel result support the proof. The work also settles an interpolation question.
Catalog record
| CREATOR | Lucas Davron, and Pierre Lissy |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 115 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-07326
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exact output tracking for the one-dimensional heat equation and applications to the interpolation problem in Gevrey classes of order 2 Lucas Davron, and Pierre Lissy Abstract Abstract This paper provides a complete characterization of the Dirichlet boundary outputs that can be exactly tracked in the one-dimensional heat equation with Neumann boundary control. The problem consists in describing the set of boundary traces generated by square-integrable controls over a finite or infinite time horizon. We show that these outputs form a precise functional space related to Gevrey regularity of order 2. In the infinite-time case, the trackable outputs are precisely those functions whose successive derivatives satisfy a weighted summability condition, which corresponds to specific Gevrey classes. For finite-time horizons, an additional compatibility condition involving the reachable space of the system provides a full characterization. The analysis relies on Fourier-Laplace transform, properties of Hardy spaces, the flatness method, and a new Plancherel-type theorem for Hilbert spaces of Gevrey functions. Beyond control theory, our results yield an optimal solution to the classical interpolation problem in Gevrey- 2 classes, which improves results of Mitjagin on the optimal loss factor. The techniques developed here also extend to variants of the heat system with different boundary conditions or observation points. Keywords Heat equation; boundary control; exact output tracking; Gevrey regularity; interpolation problem. MSC 2020 26E10, 93C20, 35K05, 46EE10. 1 Introduction 1.1 Output tracking Let us consider the one-dimensional heat problem with Neumann boundary conditions \left\{\begin{array}[]{rclcc}z_{t}(t,x)&=&z_{xx}(t,x),&0<x<1,&t>0,\\ z_{x}(t,1)&=&u(t),\\ z_{x}(t,0)&=&0,\\ z(0,x)&=&0,\\ y(t)&=&z(t,0),\end{array}\right. (1) where z=z(t,x) denotes the unknown state, u=u(t) the control, and y=y(t) is what we call the output signal . In the present article, we aim to obtain a sharp characterization of these outputs y when u ranges over L^{2}(0,T) . In other words, we look for a description of the set \mathcal{Y}(0,T):=\{y\in L^{1}_{\operatorname{loc}}[0,T):u\in L^{2}(0,T)\}, where T\in(0,\infty] is the horizon time. We refer to this as an exact output tracking problem. In general, output tracking is a central problem in automatic control theory. From both theoretical and practical perspectives, the full state of a control system is often neither directly controllable nor observable, making full-state tracking unrealistic. In such situations, one seeks instead to track only a given output of the system, such as, a boundary trace, for instance, as in the present example. Furthermore, output tracking can be viewed as a preliminary step towards the more demanding problem of output regulation , which also involves disturbance rejection and robustness considerations. Most of the existing literature deals with asymptotic tracking achieved through (possibly dynamic) feedback laws, where the controlled output asymptotically approaches a reference trajectory as t\to+\infty (see, for instance, [ 34 , Chapter 12] for related results on the one-dimensional heat equation). In contrast, much less is known about exact or approximate output tracking achieved through open-loop controls, both for finite-dimensional linear systems and for controlled partial differential equations. Some recent progress in this finite-dimensional setting was reported in [ 57 ] . In the context of the wave equation, exact output tracking in a closely related framework was investigated in [ 51 , 12 ] . Finally, for higher-dimensional generalizations of system ( 1 ), the weaker notion of approximate output tracking —where the desired tracking property is achieved up to an arbitrarily small error using an open-loop control—has been addressed in [ 1 ] . Concerning the exact tracking of ( 1 ), only a few results already exist in the literature, each in the case of finite T . On the one hand, any y\in\mathcal{Y}(0,T) satisfies y\in C^{\infty}[0,T] together with the two conditions y^{(k)}(0)=0,\quad\forall k\in\mathbb{N}, (2) and \exists C,R>0,\quad\forall t\in[0,T],\quad\forall k\in\mathbb{N},\quad|y^{(k)}(t)|\leq C\frac{(2k)!}{R^{2k}}. (3) Such a regularity result can be deduced from Gevrey’s estimates [ 23 ] ; see also [ 13 , Proposition 12.1, Chapter 5] . Moreover, in the above, the parameter R is uniform with respect to T and y . On the other hand, it has been established by Martin Rosier and Rouchon in [ 38 , Proposition 3.1] (see also Hölmgren [ 26 ] for connected results) using the flat-output method that a sufficient condition for a signal \varphi\in C^{\infty}[0,T] to belong to \mathcal{Y}(0,T) is that there is some R>1 such that \varphi satisfies ( 2 ) together with ( 3 ). Let us denote by G^{2,R}[0,T] the set of those \varphi\in C^{\infty}[0,T] satisfying the estimate ( 3 ), and G^{2,R}_{(0)}[0,T] the set of those \varphi\in G^{2,R}[0,T] which satisfy in addition ( 2 ). The above two results can be summed-up as \exists R\leqslant 1,\quad\forall\epsilon,T>0,\quad G^{2,1+\epsilon}_{(0)}[0,T]\subset\mathcal{Y}(0,T)\subset G^{2,R}_{(0)}[0,T]. Our description of \mathcal{Y}(0,T) will use Hilbert spaces of Gevrey functions similar to G^{2,R} . We start with the case T=+\infty , which is our first main result. Theorem 1 . The set \mathcal{Y}(0,\infty) consists of those \varphi\in C^{\infty}_{(0)}[0,\infty) such that \sum_{k=0}^{\infty}\left(\frac{||{\varphi^{(k+1)}}||_{L^{2}(0,\infty)}}{(2k)!2^{k}(1+k)^{3/4}}\right)^{2}<\infty. (4) The estimate ( 4 ) essentially means that \varphi is Gevrey of order 2 and radius 1/\sqrt{2} . The constant 1/\sqrt{2} is no coincidence, and could have been guessed from the characterization of the reachable space of the heat equation and the flat-output method. We introduce these two theories below. The reachable space for ( 1 ) is defined by \mathcal{R}_{T}:=\{z(T,\cdot):u\in L^{2}(0,T)\},\quad(T<\infty). (5) The set \mathcal{R}_{T} has been studied in great detail in [ 17 , 25 , 52 , 38 , 29 , 11 , 37 ] where an explicit characterization was obtained. As shown in the cited references, the set \mathcal{R}_{T} does not depend on T ; we denote it by \mathcal{R} . Moreover, it consists of those functions f that are holomorphic and even on the tilted square \Omega:=\{\zeta=a+ib\in\mathbb{C}:|a|+|b|<1\}, and such that, moreover, f^{\prime}\in L^{2}(\Omega) , where f^{\prime}=\partial f/\partial\zeta is the complex derivative. Note that such an f satisfies \frac{\partial f}{\partial{\bar{\zeta}}}=0,\quad\frac{\partial f}{\partial\zeta}\in L^{2}(\Omega), hence, regarding f as a function of two real variables we have \nabla f\in L^{2}(\Omega) . Therefore, by standard procedures (see e.g. [ 40 , §1.1.11] ), we deduce that f belongs to the Sobolev space W^{1,2}(\Omega) 1 1 1 We do not use the notation H^{1}(\Omega) for the Sobolev space W^{1,2}(\Omega) in order to save the letter H for the Hardy space H^{2} . , where \Omega is seen as a subset of \mathbb{R}^{2} . Thus, \mathcal{R}=\{f\in\operatorname{Hol}(\Omega)\cap W^{1,2}(\Omega),\,f\mbox{ even}\}, (6) where \operatorname{Hol}(D) stands for the set of holomorphic functions in D , whenever D is an open subset of \mathbb{C} . Now take a solution z of ( 1 ). It is in C^{\infty}((0,T)\times(0,1)) , and can be extended as an even and C^{\infty} function on (0,T)\times(-1,1) , which solves the heat equation therein. The flat-output method is a general procedure to parameterize the state of a system as a power series of the successive time derivatives of a well-chosen output, we refer to [ 18 ] for a general overview. For the heat equation ( 1 ), this method has been implemented by developing z as power series of x , we refer to [ 26 , 35 , 38 ] . Notably, for any u\in L^{2}(0,T) , the solution z of ( 1 ) writes z(t,x)=\sum_{k=0}^{\infty}y^{(k)}(t)\frac{x^{2k}}{(2k)!}, (7) where for fixed 0\leq t\leq T , the power series converges for |x| small enough. From ( 7 ) and Cauchy’s inequalities, we see that the largest R such that y\in G^{2,R}[0,T] is