Table of Contents
Fetching ...

Control of the half-heat equation

Andreas Hartman, Armand Koenig

TL;DR

This work analyzes null-controllability for the half-heat equation on the unit circle with controls supported on a sub-arc {\omega}. It develops a complex-analytic framework, connecting the L^2 and H^2 settings via the Riesz projection and Bergman/Hardy spaces, and establishes a time-independent characterization of null-controllable states in the H^2 setting through an observability inequality on Bergman spaces over a geometry {\Omega_T}. A key achievement is expressing necessary and sufficient conditions in terms of boundary data supported on {\omega}, including a pseudo-Carleson-measure criterion, and linking these to a Poisson-kernel/Stokes-argument formulation. The L^2 problem is then related to the H^2 results by a frequency-separation argument, yielding density results for null-controllable data and a precise correspondence between NC_{L^2}(ω) and joint NC_{H^2}(ω) conditions. Overall, the paper blends complex analysis, Bergman/Hardy space theory, and control theory to characterize reachable/steerable states for the half-heat equation, with implications for fractional and Baouendi–Grushin-type controllability via similar analytic tools.

Abstract

In this paper we investigate null-controllable initial states of the half heat equation controlled from a sub-arc $ω$ of the unit circle. We also study the projection on positive frequencies of the half-heat equation. For this projected half-heat equation, we obtain necessary as well as sufficient conditions for an initial condition to be null-controllable. These conditions, which are almost sharp, are expressed in term of projections on positive frequencies of functions supported on $ω$. From these results, and with the help of classical results on sum of holomorphic and anti-holomorphic functions, we also treat the (unprojected) half-heat equation. Surprisingly, without using our conditions on null-controllable states, we are able to show that the space of null-controllable functions does not depend on time by using a result of separation of singularities for holomorphic functions.

Control of the half-heat equation

TL;DR

This work analyzes null-controllability for the half-heat equation on the unit circle with controls supported on a sub-arc {\omega}. It develops a complex-analytic framework, connecting the L^2 and H^2 settings via the Riesz projection and Bergman/Hardy spaces, and establishes a time-independent characterization of null-controllable states in the H^2 setting through an observability inequality on Bergman spaces over a geometry {\Omega_T}. A key achievement is expressing necessary and sufficient conditions in terms of boundary data supported on {\omega}, including a pseudo-Carleson-measure criterion, and linking these to a Poisson-kernel/Stokes-argument formulation. The L^2 problem is then related to the H^2 results by a frequency-separation argument, yielding density results for null-controllable data and a precise correspondence between NC_{L^2}(ω) and joint NC_{H^2}(ω) conditions. Overall, the paper blends complex analysis, Bergman/Hardy space theory, and control theory to characterize reachable/steerable states for the half-heat equation, with implications for fractional and Baouendi–Grushin-type controllability via similar analytic tools.

Abstract

In this paper we investigate null-controllable initial states of the half heat equation controlled from a sub-arc of the unit circle. We also study the projection on positive frequencies of the half-heat equation. For this projected half-heat equation, we obtain necessary as well as sufficient conditions for an initial condition to be null-controllable. These conditions, which are almost sharp, are expressed in term of projections on positive frequencies of functions supported on . From these results, and with the help of classical results on sum of holomorphic and anti-holomorphic functions, we also treat the (unprojected) half-heat equation. Surprisingly, without using our conditions on null-controllable states, we are able to show that the space of null-controllable functions does not depend on time by using a result of separation of singularities for holomorphic functions.

Paper Structure

This paper contains 23 sections, 36 theorems, 159 equations, 5 figures.

Key Result

Proposition 1.2

Let $\omega$ be a strict open subset of $\mathbb T$ and $T>0$. Let $\Omega_T \mathrel{\mathord{:}\mathord=} \{ z\in \mathbb C,\ 1<|z|<\mathrm e^T,\ z/|z| \in \omega\}$ (see fig-Omega). Let $f_0 \in H^2$. The following assertions are equivalent: If these assertions are satisfied, we can choose the control $u$ that steers $f_0$ to $0$ such that $\|u\|_{L^2([0,T]\times\omega)} \leq \mathrm e^T C_{f

Figures (5)

  • Figure 1: Illustration of $\Omega_T^*$ (yellow partial ring inside of ${\mathbb{D}}$) and $\Omega_T$ (red partial ring outside ${\mathbb{D}}$).
  • Figure 2: A plot of $\theta\mapsto P_+g(\mathrm e^{\mathrm i\theta})$ for $\theta_0 = \pi/4$. The real part is in blue, the imaginary part in green.
  • Figure 3: Illustration of $\widehat{\Omega}_{T'}$ and $\Omega_T$ when $T'<T$. The partial ring $\Omega_T$ is in red (including the reddish-yellow part). $\widehat{\Omega}_{T'}$ is in yellow (also including the reddish-yellow part). The intersection of $\Omega_T$ and $\widehat{\Omega}_{T'}$ is $\Omega_{T'}$ colored in reddish-yellow.
  • Figure 4: Left figure: illustration of $\widetilde{\Omega}_T$ (in green), with $\Omega_T$ in red, the paths $\Gamma_i$ ($i=0,1,2$) and the points $\zeta_i'$ and $\mathrm e^{\eta_0}\zeta'_i$ ($i=1,2$). Right figure: illustration of $\widetilde{\Omega}_{T,\eta}$ (in green) and the path $\omega_\eta$.
  • Figure 5: Illustration of the paths $\gamma_\epsilon$ and $\gamma_+$. They are used in the proof of \ref{['th-lemma-ortho']} to prove with Morera's theorem that the function $H(z) = f(z)$ if $|z|<1$ and $f(z) = h(z)$ if $|z|\geq 1$ and $z\in\notin(-\infty,0]$ is holomorphic.

Theorems & Definitions (70)

  • Definition 1.1
  • Proposition 1.2
  • Theorem 1.3
  • Theorem 1.4
  • Theorem 1.5
  • Corollary 1.6
  • Theorem 1.7
  • Corollary 1.8
  • Corollary 1.9
  • Theorem 1.10
  • ...and 60 more