Abstract
In this article we study the semiclassical resolvent estimate for the non-selfadjoint Baouendi–Grushin operator on the two-dimensional torus \({\mathbb {T}}^2={\mathbb {R}}^2/(2\pi {\mathbb {Z}})^2\) with Hölder dampings. The operator is subelliptic degenerating along the vertical direction at \(x=0\). We exhibit three different situations: (i) the damping region verifies the geometric control condition with respect to both the non-degenerate Hamiltonian flow and the vertical subelliptic flow; (ii) the undamped region contains a horizontal strip; (iii) the undamped part is a horizontal line. In all of these situations, we obtain sharp resolvent estimates. Consequently, we prove the optimal energy decay rate for the associated damped waved equations. For (i) and (iii), our results are in sharp contrast to the Laplace resolvent since the optimal bound is governed by the quasimodes in the subelliptic regime. While for (ii), the optimality is governed by the quasimodes in the elliptic regime, and the optimal energy decay rate is the same as for the classical damped wave equation on \({\mathbb {T}}^2\). Our analysis contains the study of adapted two-microlocal semiclassical measures, construction of quasimodes and refined Birkhoff normal-form reductions in different regions of the phase-space. Of independent interest, we also obtain the propagation theorem for semiclassical measures of quasimodes microlocalized in the subelliptic regime.
Similar content being viewed by others
Notes
Here the polynomial rate means a rate like \(t^{-p}\) for some \(p>0\), not necessary an integer. We respect this terminology from the existing literature.
We will recall the precise statement in “Appendix D”.
For the specific operator \({\mathcal {A}}\), the condition \(i{\mathbb {R}}\cap \sigma (\dot{{\mathcal {A}}})=\emptyset \) is satisfied as a consequence of the unique continuation property, see [38].
We will require enough regularity on b to use pseudodifferential symbolic calculus when necessary along the proof.
For the bound \(O(h^{-3})\) it is sufficient that \(b \in L^\infty \).
The argument consists of applying Corollary A.7 to \(\sqrt{{\textbf{a}}_{h,\epsilon ,{\mathbb {R}}}+\delta }\), taking the triple limit and finally letting \(\delta \rightarrow 0\).
Here the point is that \(\chi _{\epsilon }(h\eta ){\widetilde{a}}_R^h(y,\eta )\) is independent of x and \(\xi \) variable, hence the dangerous derivative in x does not appear in the formal symbolic product \(\chi _{\epsilon }(h\eta )\widetilde{{\textbf{a}}}_R^h(y,\eta ,\eta ')_{\#}\chi _R^2(x^2\eta ^2+\xi ^2-1)\).
Here we only encounter compositions and commutators between a scalar-valued p.d.o (in \((y,\eta )\) variables) and a compact operator-valued p.d.o.
By the periodic extension procedure described in Appendix A, it suffices to compute the commutator for operators acting on \({\mathcal {D}}'({\mathbb {R}}^2)\) instead of \({\mathcal {D}}'(M_0)\).
Here we can choose any cutoff with support away from zero, but to sake the notation, we make the choice of cutoff \(\chi _1\).
As usual, here we drop the subindex for parameters.
The integral \(\int _{{\mathbb {R}}^8}e^{i\langle QZ,Z\rangle }dZ\) should be understood as an oscillatory integral, in the sense that
$$\begin{aligned} \int _{{\mathbb {R}}^8}e^{i\langle QZ,Z\rangle }dZ=\lim _{\epsilon \rightarrow 0^+}\int _{{\mathbb {R}}^8}e^{i\langle QZ,Z\rangle -\epsilon |Z|^2}. \end{aligned}$$
References
Anantharaman, N., Fermanian-Kammerer, C., Macià, F.: Semiclassical completely integrable systems: long-time dynamics and observability via two-microlocal wigner measures. Am. J. Math. 137(3), 57–638 (2015)
Anantharaman, N., Léautaud, M.: Sharp polynomial decay rates for the damped wave equation on the torus. Anal. PDE 7(1), 159–214 (2014)
Anantharaman, N., Léautaud, M., Macià, F.: Wigner measures and observability for the Schrödinger equation on the disk. Invent. Math. 206(2), 485–599 (2016)
Anantharaman, N., Léautaud, M., Macià, F.: Delocalization of quasimodes in the disk. Comp. Rend. Math. e. 354(3), 256–263 (2016)
Anantharaman, N., Macià, F.: Semiclassical measures for the Schrödinger equation on the torus. J. Eur. Math. Soc. (JEMS) 16(6), 1253–1288 (2014)
Arnaiz, V.: Construction of quasimodes for non-selfadjoint operators via propagation of Hagedorn wave-packets. J. Spectr. Theory 12(2), 745–812 (2022)
Arnaiz, V., Rivière, G.: Semiclassical asymptotics for nonselfadjoint harmonic oscillators. Pure Appl. Anal. 2(2), 427–445 (2020)
Asch, M., Lebeau, G.: The spectrum of the damped wave operator for a bounded domain in \({ R}^2\). Exp. Math. 12(2), 227–241 (2003)
Bardos, C., Lebeau, G., Rauch, J.: Sharp sufficient conditions for the observation, control, and stabilization of waves from the boundary. SIAM J. Control. Optim. 30(5), 1024–1065 (1992)
Beauchard, K., Cannarsa, P.: Heat equation on the Heisenberg group: observability and applications. J. Differ. Equ. 262(8), 4475–4521 (2017)
Beauchard, K., Cannarsa, P., Guglielmi, R.: Null controllability of Grushin type operators in dimension two. J. Eur. Math. Soc. (JEMS) 16(1), 67–101 (2014)
Beauchard, K., Dardé, J., Ervedoza, S.: Minimal time issues for the observability of Grushin-type equations. Ann. Inst. Fourier (Grenoble) 70(1), 247–312 (2020)
Beauchard, K., Pravda-Starov, K.: Null-controllability of hypoelliptic quadratic differential equations. J. Éc. Polytech. Math. 5, 1–43 (2018)
Borichev, A., Tomilov, Y.: Optimal polynomial decay of functions and operator semigroups. Math. Ann. 347(2), 455–478 (2010)
Burq, N.: Decays for Kelvin–Voigt damped wave equations I: the black box perturbative method. SIAM J. Control. Optim. 58(4), 1893–1905 (2020)
Burq, N., Gérard, P.: Condition nécessaire et suffisante pour la contrôlabilité exacte des ondes. C. R. Acad. Sci. Paris Sér. I Math. 325(7), 749–752 (1997)
Burq, N., Gérard, P.: Stabilization of wave equations on the torus with rough dampings. Pure Appl. Anal. 2(3), 627–658 (2020)
Burq, N., Hitrik, M.: Energy decay for damped wave equations on partially rectangular domains. Math. Res. Lett. 14(1), 35–47 (2007)
Burq, N., Sun, C.: Time optimal observability for the Grushin–Schrödinger equation. Anal. PDE 15(6), 1487–1530 (2022)
Burq, N., Zworski, M.: Control in the presence of a black box. J. Am. Math. Soc. 17, 443–471 (2004)
Chaves-Silva, F., Sun, C.: On the stabilization of a hyperbolic Stokes system under geometric control condition. Z. Angew. Math. Phys. 71(4), 24 (2020)
Colin de Verdière, Y., Hillairet, L., Trélat, E.: Spectral asymptotics for sub-Riemannian Laplacians I: quantum ergodicity and quantum limits in the 3-dimensional contact case. Duke Math. J. 167(1), 109–174 (2018)
Cox, S., Zuazua, E.: The rate at which the energy decays in a damped string. Commun. Partial Differ. Equ. 19(1–2), 213–243 (1994)
Datchev, K., Kleinhenz, P.: Sharp polynomial decay rates for the damped wave equation with Hölder-like damping. Proc. Am. Math. Soc. 148(8), 3417–3425 (2020)
Dencker, N., Sjöstrand, J., Zworski, M.: Pseudospectra of semiclassical (pseudo-) differential operators. Commun. Pure Appl. Math. 57(3), 384–415 (2004)
Duprez, M., Koenig, A.: Control of the Grushin equation: non-rectangular control region and minimal time. ESAIM Control Optim. Calc. Var. 26, 3 (2020)
Fermanian-Kammerer, C.: Measures semi-classiques 2-microlocales. C. R. Acad. Sci. Paris Sér. I Math. 331(7), 515–518 (2000)
Fermanian-Kammerer, C., Letrouit, C.: Observability and controllability for the Schrödinger equation on quotients of groups of Heisenberg type. J. Éc. Polytech. Math. 8, 1459–1513 (2021)
Gérard, P.: Microlocal defect measures. Commun. Partial Differ. Equ. 16(11), 1761–1794 (1991)
Hwang, I.L.: The \(L^2\)-Boundedness of Pseudo-differential operators. Trans. Am. Math. Soc. 302(1), 55–76 (1987)
Kleinhenz, P.: Stabiliation rates for the damped wave equation with Hölder-regular damping. Commun. Math. Phys. 369(3), 1187–1205 (2019)
Koenig, A.: Non-null-controllability of the Grushin operator in 2D. C. R. Math. Acad. Sci. Paris 355(12), 1215–1235 (2017)
Laurent, C., Léautaud, M.: Logarithmic decay for damped hypoelliptic wave and Schrödinger equations. SIAM J. Control Optim. 59(3), 1881–1902 (2021)
Léautaud, M., Lerner, N.: Energy decay for a locally undamped wave equation. Ann. Fac. Sci. Toulouse Math. (6) 26(1), 157–205 (2017)
Lebeau, G.: Equation des ondes amorties, Algebraic and geometric methods in mathematical physics (Kaciveli, 1993), Mathematical Physics Studies, vol. 19, pp. 73–109. Kluwer Academic Publishers, Dordrecht (1996)
Letrouit, C.: Subelliptic wave equations are never observable. To appear in Anal. PDEs. arXiv:2002.01259
Letrouit, C.: Exact observability properties of subelliptic wave and Schrödinger equations. Actes du séminaire de Théorie Spectrale et Géométrie
Letrouit, C., Sun, C.: Observability of Baouendi–Grushin type evolutions equations through resolvent estimates. J. Instit. Math. Jussieu. https://doi.org/10.1017/s1474748021000207
Macià, F., Rivière, G.: Concentration and non-concentration for the Schrödinger evolution on Zoll manifolds. Commun. Math. Phys. 345(3), 1019–1054 (2016)
Macià, F., Rivière, G.: Two-microlocal regularity of quasimodes on the torus. Anal. PDE 11(8), 2111–2136 (2018)
Macià, F., Rivière, G.: Observability and quantum limits for the Schrödinger equation on \(\mathbb{S}^d\). In: Probabilistic Methods in Geometry, Topology and Spectral Theory, Contemporary Mathmaticals, vol. 739, pp. 139–153. Centre Rech. Math. Proc., American Mathematics Society (2019)
Martinez, A.: An introduction to semiclassical and microlocal analysis. In: Universitext. Springer, New York, (2002). ISBN: 0-387-95344-2
Rauch, J., Taylor, M.: Exponential decay of solutions to hyperbolic equations in bounded domains. Indiana Univ. Math. J. 24, 79–86 (1974)
Sjöstrand, J.: Resolvent estimates for non-selfadjoint operators via semigroups. In: Around the Research of Vladimir Maz’ya. III, pp. 359–384, Int. Math. Ser. (N. Y.) 13, Springer, New York (2010)
Sun, C.: Sharp decay rate for the damped wave equations with convex-shaped damping. Int. Math. Res. Notices rnac022 https://doi.org/10.1093/imrn/rnac022
Zworski, M.: Semiclassical analysis. In: Graduate Studies in Mathematics, vol. 138. American Mathematical Society, Providence, RI (2012), ISBN: 978-0-8218-8320-4
Acknowledgements
The first author has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Grant agreement No. 725967), and has been partially supported by Grant MTM2017-85934-C3-3-P (MINECO, Spain). The second author is partially supported by the ANR Grant ODA (ANR-18-CE40-0020-01) and he would like to thank ICERM of Brown University for the kind hospitality, where the end of this research was carried out. We also thank Gabriel Rivière for his comments on a draft version of this article. Finally, we are grateful for anonymous referees.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors have no competing interests to declare that are relevant to the content of this article.
Additional information
Communicated by S. Dyatlov.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: Special Symbol Classes and Quantizations
In this section, we will collect some special symbolic calculus needed in Sect. 3. First, let us recall that a symbol \({\textbf{a}}\in S^{\rho ,\delta }({\mathbb {R}}^d)\) is a smooth function on \({\mathbb {R}}_{z,\zeta }^{2d}\) such that
For a symbol \({\textbf{a}}\), we define the Weyl quantization \(\textrm{Op}_{1,{\mathbb {R}}^d}^w({\textbf{a}})\) via the formula
for any Schwartz function \(f\in {\mathcal {S}}({\mathbb {R}}^d)\). When \(z=(x,y)\in {\mathbb {R}}^{2m},\zeta =(\xi ,\eta )\in {\mathbb {R}}^{2m}\), we denote by \( {\text {Op}}_{1}^{\textrm{w},(x,\xi )}({\textbf{a}}),{\text {Op}}_1^{\textrm{w},(y,\eta )}({\textbf{a}}) \) the partial quantization of \({\textbf{a}}\) with respect to \((x,\xi )\) and \((y,\eta )\) variables, respectively. Note that we can write
This means that we quantize the \({\mathcal {L}}(L^2({\mathbb {R}}_x^n))\)-valued symbol \(\textrm{Op}_1^{\textrm{w},(x,\xi )}({\textbf{a}})={\textbf{a}}^{\textrm{w}}(x,D_x;y,\eta )\) with domain \((y,\eta )\in {\mathbb {R}}^{2n}\).
Recall the following version of the Calderon–Vaillancourt theorem for \(S^{0,0}\) symbols:
Proposition A.1
[30, Theorem 2]. Let \({\textbf{a}}\in S^{0,0}({\mathbb {R}}^{d})\). Then
where \(C_0>0\) is a uniform constant.
By rescaling \(u(z)\mapsto {\widetilde{u}}({\widetilde{z}}):=h^{\frac{d}{4}}u(h^{\frac{1}{2}}{\widetilde{z}})\), we obtain that
where \({\textbf{a}}_h({\widetilde{z}},{\widetilde{\zeta }})={\textbf{a}}(h^{\frac{1}{2}}{\widetilde{z}},h^{\frac{1}{2}}{\widetilde{\zeta }})\). Consequently, we have
Corollary A.2
(Theorem 5.1, [46]). Suppose that \({\textbf{a}}\in S^{0,0}({\mathbb {R}}^d)\). Then
where \(C_0\) is an absolute constant.
1.1 A.1 Explicit quantization on \(M_0={\mathbb {R}}_x\times {\mathbb {T}}_y\)
In this subsection, we follow the procedure of Chapter 5, Section 3 of [46] to define quantization for partially periodic symbols. Denote by \(\iota : L^2(M_0)\rightarrow {\mathcal {S}}'({\mathbb {R}}^2)\) the identification of a function in \(L^2(M_0)\) as a tempered distribution in \({\mathcal {S}}'({\mathbb {R}}^2)\). In coordinate, we can write
where \({\textbf{1}}_{{\mathbb {T}}_y}\) stands for the restriction to the fundamental domain \((-\pi ,\pi )\) of \({\mathbb {T}}_y\). Denote by \({\mathcal {S}}_{\text {per}}'({\mathbb {R}}^2)\) the subspace of partially \(2\pi \)-periodic distributions (in y variable) which is the image of \(\iota \). Note that
Symbols \({\textbf{a}}\) on \(T^*M_0\) can be identified as partially periodic symbols on \(T^*{\mathbb {R}}^2\), namely \({\textbf{a}}\in C^{\infty }({\mathbb {R}}^4)\), bounded as well as its derivatives and satisfying
The quantization \(\textrm{Op}_1^\textrm{w}(\cdot )\) on \(M_0\) is naturally defined by
where the notation \({\text {Op}}_{1,{\mathbb {R}}^2}^\textrm{w}\) is to emphasize the quantization is on \({\mathbb {R}}^2\), given by the formula (A.1).
The quantization formula (A.3) can be expanded explicitly as
where
or equivalently,
where \(\tau _{y_0}f(x,y)=f(x,y-y_0)\) is the translation operator. If there exist \(\sigma \ge 0\), \(M\ge 2\) such that some h-dependent family of symbols \({\textbf{a}}_h\) satisfies additionally \(|\partial _{\eta }^{m}{\textbf{a}}_h|\lesssim h^{m\sigma }\) for any \(m\le M\) (the situation of \(\sigma >0\) appears when considering the semiclassical quantization), then in \(\textrm{Op}_1^\textrm{w}({\textbf{a}}_h)\) we may only consider the contribution \({\textbf{A}}_{k}\) for \(|k|\le 1\). Indeed, for \(|k|\ge 2\), we have, for any \(y,y'\in {\mathbb {T}}\), \(|y-y'+2\pi k|\ge (|k|-1)2\pi \). By writing \(\textrm{e}^{i(y-y'+2\pi k)}\) as \(\frac{\partial _{\eta }^m(\textrm{e}^{i(y-y'+2\pi k)} )}{(i(y-y'+2\pi k))^m }\) and using integration by parts, we deduce that
Consequently, we have the following version of the Calderon–Vaillancourt theorem:
Corollary A.3
(Calderon–Vaillancourt). There exists \(C_0>0\), such that for all \(a\in C^{\infty }(T^*M_0)\),
1.2 A.2 Special symbol classes associated to the second microlocalization
Definition A.4
Assume that \(h,\epsilon ,R\) are sequences of parametersFootnote 13 satisfying the following asymptotic:
The \((h,\epsilon ,R)\) parameter-dependent symbol class \({\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^6)\) consists of families of smooth functions \({\textbf{a}}(x,x_1,y,y_1,\xi ,\eta ;h,\epsilon ,R)\), satisfying the following hypotheses:
-
(i)
There exists \(K>0\) such that uniformly in parameters \(h,\epsilon ,R\), \({\textbf{a}}(x,x_1,y,y_1,\xi ,\eta ;h,\epsilon ,R)\equiv 0\) when
$$\begin{aligned} \sqrt{(x+x_1)^2+(y+y_1)^2}> K\quad \text {or}\quad \sqrt{\xi ^2+\eta ^2}> Kh^{-1}. \end{aligned}$$ -
(ii)
For any \(k,k_1,m,m_1,l,j\), there exists \(C_{k,k_1,m,m_1,l,j}>0\), such that for all \(h,\epsilon ,R\), the following estimate holds
$$\begin{aligned} \sup _{{\mathbb {R}}^6}|\partial _x^k\partial _{x_1}^{k_1}\partial _y^m\partial _{y_1}^{m_1}\partial _{\xi }^l\partial _{\eta }^j{\textbf{a}}|\le C_{k,k_1,m,m_1,l,j}\cdot \big (\frac{\epsilon }{h}\big )^{k+k_1}\cdot h^{l+j}. \end{aligned}$$(A.6)
Similarly, \({\textbf{a}}={\textbf{a}}(x,y,\xi ,\eta ;h,\epsilon ,R)\) belongs to \({\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\) if the analogue of the hypotheses \(\mathrm {(i)},\mathrm {(ii)}\) hold without variables \(x_1,y_1\).
When there is no risk of confusing, we will not display the dependence in \(h,\epsilon ,R\) explicitly for symbols in \({\textbf{S}}_{1,1;1^-,0}^0\). Moreover, when the symbol \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\) is \(2\pi \)-periodic in y variable, we denote by \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0(T^*M_0)\). We have the following \(L^2\) boundedness property for the quantization of this class of symbols:
Proposition A.5
Assume that \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^6)\) and let \(T_{{\textbf{a}}}\) be the linear operator on \({\mathcal {S}}'({\mathbb {R}}^2)\) with the Schwartz kernel
where \(x,x_1,y,y_1\in {\mathbb {R}}\). Then \(T_{{\textbf{a}}}\in {\mathcal {L}}(L^2({\mathbb {R}}^2))\) uniformly in \(h,\epsilon \) and R obeying the asymptotic (A.6). More precisely,
where the first constant \(C_0\) is independent of \({\textbf{a}}\). Consequently, if \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\) and is \(2\pi \)-periodic in y, then
Proof
Consider the scalings \(x=\frac{h}{\epsilon ^{1/2}}X, y=h^{1/2}Y\) and \(\xi =\frac{\epsilon ^{1/2}}{h}\Xi , \eta =h^{-1/2}\Theta \) and for \(f,g\in L^2({\mathbb {R}}^2)\), we denote by
Note that we have \(\Vert f\Vert _{L^2({\mathbb {R}}^2)}=\Vert F\Vert _{L^2({\mathbb {R}}^2)}\) and \(\Vert g\Vert _{L^2({\mathbb {R}}^2)}=\Vert G\Vert _{L^2({\mathbb {R}}^2)}\). Direct computation yields
where
with \(X_1=\frac{\epsilon ^{1/2}}{h}x_1, Y_1=h^{-1/2}y_1\). By the assumption on \({\textbf{a}}\), we verify that
Then by a variant version of the Calderón–Vaillancourt theorem (Theorem 2.8.1 of [42]), we deduce that \(T_{\widetilde{{\textbf{a}}}}\) are uniformly bounded on \(L^2({\mathbb {R}}^2)\). This means that
where the first constant \(C_0\) does not depend on \({\textbf{a}}\). By (A.8) and duality, we obtain the desired bound for \(\Vert T_{{\textbf{a}}}\Vert _{{\mathcal {L}}(L^2({\mathbb {R}}^2))}\). For \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\), \(2\pi \)-periodic in y, the extension to \({\text {Op}}_1^\textrm{w}({\textbf{a}})\) on \(L^2(M_0)\) follows simply from the decomposition (A.4) and (A.5). The proof of Proposition A.5 is now complete. \(\square \)
Next, we need the following Lemma to derive a G\(\mathring{\textrm{a}}\)rding type inequality for symbols in \({\textbf{S}}_{1,1;1^-,0}^0\):
Lemma A.6
Let \({\textbf{a}},{\textbf{b}}\in {\textbf{S}}_{1,1;1^-,0}^0(T^*M_0)\). Then there exists \(C_1>0\), such that uniformly in \(h,\epsilon ,R\) satisfying the asymptotic (A.6),
Proof
We first prove the estimate for \({\textbf{a}},{\textbf{b}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\). Recall that
with
where
With \(X=(z,\zeta ), X_j=(z_j,\zeta _j)\in {\mathbb {R}}^4\), \(j=1,2\), \(Z=(X_1,X_2)\in {\mathbb {R}}^8\), we can write
where \({\mathcal {A}}(X,Z)={\textbf{a}}(X+X_1){\textbf{b}}(X+X_2)\), and
Note that \(Q^{-1}=Q\) and \(\langle QZ,Z\rangle =-2\sigma (X_1,X_2)\). Direct computation yields
where \({\mathcal {D}}_Z=D_{\zeta _1}\cdot D_{z_2}-D_{z_1}\cdot D_{\zeta _2}\). Indeed, sinceFootnote 14
by Taylor expansion
and the simple observation that \(Z=\frac{Q^{-1}}{2i}\nabla _Z(\textrm{e}^{i\langle QZ,Z\rangle })\), we have
Noting that \(Q^{-1}=-Q\) and \(\nabla _Z\cdot Q\nabla _Z=-2{\mathcal {D}}_Z\), we obtain (A.9).
Note that \({\mathcal {D}}_Z{\mathcal {A}}(X,tZ)\) is a linear combination of the terms
with \(\widetilde{{\textbf{a}}},\widetilde{{\textbf{b}}}\in {\textbf{S}}_{1,1;1^-,0}^0({\mathbb {R}}^4)\). It suffices to show that for \(t\in (0,1]\),
Since the derivatives on \(z,\zeta \) will fall on \(\widetilde{{\textbf{a}}},\widetilde{{\textbf{b}}}\), by definition of the symbol class, we only need to show that
uniformly in \(h,\epsilon ,R\) and \(t\in (0,1]\). Denote by
the partial Fourier transform, then we have
Since \(\widetilde{{\textbf{a}}},\widetilde{{\textbf{b}}}\in {\textbf{S}}_{1,1;1^-,0}^0\), it follows from the integration by part that
for all \(N\ge 1\) and \(|\theta |\ge 1\). This leads to (A.10). Once we have established (A.10), applying Proposition A.5, we obtain that
Next we extend the above estimate to symbols that are \(2\pi \)- periodic in y. Take \(f,g\in L^2(M_0)\), we have
where we used (A.2). By self-adjointness, the above quantity equals to
Recall that \(\iota =\sum _{k\in {\mathbb {Z}}}\tau _{2\pi k}({\textbf{1}}_{{\mathbb {T}}_y}\cdot )\), from the discussion below (A.5), we have for all \(|k|\ge 2\),
When \(|k|\le 1\), by (A.11),
Adding up all \(k\in {\mathbb {Z}}\), we obtain that
By duality, the desired result then follows and we complete the proof of Lemma A.6. \(\square \)
Consequently, we have:
Corollary A.7
Let \({\textbf{a}}\in {\textbf{S}}_{1,1;1^-,0}^0(T^*M_0)\), then there exists \(C_1>0\), such that
Appendix B: Exact Computations for Commutators
In this part, we collect some computations for commutators needed in Sect. 3. Below, all computations will be done only for operators acting on \(L^2({\mathbb {R}}^2)\), since all the commutators are of the form \([P,{\text {Op}}_1^\textrm{w}(q)]\), with some differential operator P and a symbol q that is periodic in y, thanks to the identification (A.3). First we prove the commutator formula below, used to prove the propagation theorem for the scalar-valued second semiclassical microlocal measure:
Lemma B.1
Let \(q(z,\zeta )\in {\mathcal {S}}({\mathbb {R}}^4)\). We have the following formula
where \(V_1(x)=V(x)-x^2=O(x^3)\) and the operator \({\mathcal {R}}_h(q)\) has the Schwartz kernel
with
and
Remark .1
In our applications of the formula, apart from the first operator on the right hand side, the others are all viewed as remainders.
Proof
The proof follows from a direct computation. Write
we have for any \(f\in {\mathcal {S}}({\mathbb {R}}^2)\),
After arrangement, we have
where \({\mathcal {T}}_1,{\mathcal {T}}_2\) have Schwartz kernels
Since \(V(x)=x^2+V_1(x)\), \(V_1(x)=O(|x|^3)\), we have
We can further write
hence
where \({\mathcal {R}}_1\) has the Schwartz kernel
For \({\mathcal {T}}_2\), by Taylor expansion and integration by part, we can write its kernel as
which is part of the remainder. We complete the proof of Lemma B.1. \(\square \)
The following formula will be used to prove the propagation formula for the operator-valued second semiclassical microlocal measures:
Lemma B.2
Let \(q(z,\zeta )\in {\mathcal {S}}({\mathbb {R}}^4)\). We have the following formula
where the operator \({\mathcal {R}}_q\) has the Schwartz kernel
where
and
Proof
The proof follows from a direct computation. We write
We observe that
and the third operator \([V(hx),{\text {Op}}_1^\textrm{w}(q)]h^2D_y^2\) has the Schwartz kernel
Since \(\frac{h^2}{(2\pi )^2}\int _{{\mathbb {R}}^2}(V(hx)-V(hx'))\eta ^2q\big (\frac{z+z'}{2},\zeta \big )\textrm{e}^{i(z-z')\cdot \zeta }d\zeta \) is the Schwartz kernel of the operator
we get
where the operator \({\mathcal {T}}\) has the Schwartz kernel
or equivalently,
By Taylor expansion and doing the integration by part, we can further write
and
By organizing terms, we complete the proof of Lemma B.2. \(\square \)
To end this section, we collect an elementary averaging Lemma for the one-dimensional harmonic oscillator:
Lemma B.3
Denote by \(H_0=D_x^2+x^2\) is the harmonic oscillator and \({\mathcal {U}}(t)=\textrm{e}^{itH_0}\) is the associated propagator, then for any \(k\in {\mathbb {N}}\),
Proof
Consider the ladder operators \({\mathcal {L}}_{\pm }:=x\mp iD_x\). These operators have properties:
and \({\mathcal {L}}_+:{\mathcal {H}}_n\mapsto {\mathcal {H}}_{n+1}\) while \({\mathcal {L}}_-:{\mathcal {H}}_n\rightarrow {\mathcal {H}}_{n-1}\), where \({\mathcal {H}}_n\) is the eigenspace of \(H_0\) associated with the eigenvalue \(2n+1\). Let \(\Pi _n: L^2({\mathbb {R}})\rightarrow {\mathcal {H}}_n\) be the orthogonal projection. From this, we deduce that the averaging \(\int _0^{2\pi }{\mathcal {U}}(t)^*(\cdots ){\mathcal {U}}(t)dt\) for a monomial composed of \({\mathcal {L}}_{\pm }\) is non-zero if and only if the number of \({\mathcal {L}}_+\) equals the number of \({\mathcal {L}}_-\). Since \(x=\frac{1}{2}({\mathcal {L}}_++{\mathcal {L}}_-)\), each monomial in the expansion of \(x^{2k-1}\) has different numbers of \({\mathcal {L}}_+,{\mathcal {L}}_-\), thus (B.3) holds. The second identity (B.4) follows from the explicit expansion of \(x^2=\frac{1}{4}(2H_0+{\mathcal {L}}_+^2+{\mathcal {L}}_-^2)\). \(\square \)
Appendix C: Subelliptic a Priori Estimates
Lemma C.1
There exists \(C_0>0\) such that for all \(u\in H_G^2({\mathbb {T}}^2)\),
Proof
The inequality \(\Vert \Delta _G u \Vert _{L^2}\le \Vert u\Vert _{{\dot{H}}_G^2}\) follows trivially by the triangle inequality, hence it suffices to prove the other. Let \(\chi \) be a bump function which is equal to 1 near 0. We decompose \(u=v+w\), with
Since \((-\Delta _Gu,u)=\Vert \nabla _Gu\Vert _{L^2}^2\), we deduce that for test functions \(\varphi =1-\chi (x)\) or \(\chi (x)\),
Therefore, it suffices to show that
Note that \(\Delta _G\) is elliptic on supp\((1-\chi )\subset {\mathbb {T}}^2{\setminus }\{x=0\}\), from the support property of v, we deduce that
To estimate \(\Vert w\Vert _{{\dot{H}}_G^2}\), we expand w as Fourier series in y, i.e. \(w=\sum _{n\in {\mathbb {Z}}}w_n(x)\textrm{e}^{iny}\). It suffices to show that uniformly in n,
where \({\mathcal {L}}_n=-\partial _x^2+n^2V(x)\) and \(w_n\) are supported on supp\((\chi )\). The estimate is trivial when \(n=0\), so below we assume that \(n\ne 0\), and without loss of generality, we assume that \(n>0\). Set \(f_n={\mathcal {L}}_nw_n\) and consider the change of variable \(z=n^{\frac{1}{2}}x\) and \({\widetilde{w}}_n(z)=w_n(x), {\widetilde{f}}_n(z)=f_n(x)\), we have
where \(\widetilde{{\mathcal {L}}}_n=-\partial _z^2+nW\big (\frac{z}{\sqrt{n}}\big )^2\). By rescaling, it suffices to show that
Having in mind that \(\sqrt{n}W\big (\frac{z}{\sqrt{n}}\big )\approx z\), the desired estimate is nothing but the a priori estimate for the elliptic equation \((-\partial _z^2+z^2){\widetilde{w}}={\widetilde{g}}\).
We expand
Integration by part yields
Note that the second term containing the commutator can be bounded from below by
Plugging into (C.3) and using Young’s inequality \(AB\le \epsilon A^2+\frac{4}{\epsilon }B^2\), we deduce that
Since
this implies (C.2). The proof of Lemma C.1 is complete.
\(\square \)
Appendix D: Equivalence to the Semiclassical Resolvent Estimate
Let us recall the classical theorem of Borichev–Tomilov:
Proposition D.1
[14] The following statements are equivalent:
Denoting \(\Pi _0\) the spectral projector of \({\mathcal {A}}\) on \(\ker {\mathcal {A}}\), since \(\textrm{e}^{t {\mathcal {A}}} = \textrm{e}^{t \dot{{\mathcal {A}}}} (\textrm{Id}- \Pi _0) + \Pi _0\), we have, if (a) or (b) hold true, that the semigroup \(\textrm{e}^{t {\mathcal {A}}}\) is stable at rate \(t^{-\alpha }\). In what follows, \({\mathcal {A}}\) is given by (1.3), associated to the damping b.
Lemma D.2
For sufficiently small \(h>0\), the following resolvent estimates are equivalent:
Proof
Essentially, the proof is given in [2]. For the sake of completeness, we provide the proof here. Denote by \(U=(u,v)^t\) and \(F=(f,g)^t\), then \((ih^{-1}-{\mathcal {A}})U=F\) is equivalent to
The implication (a)\(\implies \) (b) follows from making a special choice \((f,g)=(0,g)\in H_G^1\times L^2\).
To prove (b) \(\implies \) (a), we first claim that:
-
(i)
\(\Vert (-h^2\Delta _G-1+ihb)^{-1}\Vert _{L^2\rightarrow H_G^1}\lesssim h^{-\alpha -2}\);
-
(ii)
\(\Vert (-h^2\Delta _G-1+ihb)^{-1}\Vert _{H_G^{-1}\rightarrow L^2}\lesssim h^{-\alpha -2}\).
Indeed, assume that
from the energy identity
we deduce that
The hypothesis (b) implies that \(\Vert w\Vert _{L^2}\lesssim h^{-\alpha -1}\Vert r\Vert _{L^2}\), hence \(\Vert w\Vert _{H_G^1}\lesssim h^{-\alpha -2}\Vert r\Vert _{L^2}\), and this verifies (i). Note that the argument above is also valid for \((-h^2\Delta _G-1-ihb(y))^{-1}\), which is the adjoint of \((-h^2\Delta _G-1+ihb(y))^{-1}\). By duality, we obtain (ii).
Finally, from (b) and (ii),
From the energy identity
hence
Consequently,
thanks to (D.1). Finally, from \(u=-ihv-ihf\), we deduce that \(\Vert u\Vert _{H_G^1}\lesssim h^{-\alpha }\Vert (f,g)\Vert _{H_G^1\times L^2}\). This completes the proof of Lemma D.2 .
\(\square \)
Appendix E: Averaging Method in Finite Dimension
In this part of Appendix, we prove the following well-known finite-dimensional averaging lemma, used in Sect. 5:
Lemma E.1
Let \(D \in {\mathbb {R}}^{n \times n}\) be a diagonal matrix with entries \(\lambda _1< \cdots < \lambda _n\). Let \(A_j \in \mathbb {{\mathbb {R}}}^{n\times n}\) for \(j= 1, \ldots , N\) be self-adjoint matrices. Then for all \(N\in {\mathbb {N}}\), there exist diagonal matrices \(D_j\in {\mathbb {R}}^{n\times n}\) for all \(j=1,\ldots , N\), such that for any sufficiently small \(\epsilon \), there is a unitary matrix \({\mathfrak {U}}_N(\epsilon ) \in {\mathbb {C}}^{n\times n}\), close to the identity, such that
Proof
Write \({\mathfrak {U}}_1 = \textrm{e}^{i \epsilon F_1}\) with \(F_1 \in {\mathbb {R}}^{n\times n}\) to be chosen. Then, Taylor expansion gives
We choose \(F_1\) and \(D_1\) so that
This is possible since the eigenvalues of D are distinct, by taking
where \(\delta _{j_1 j_2}\) denotes the Kronocker delta. Then we obtain:
Iterating this procedure, we obtain the claim by defining \({\mathfrak {U}}_N := \textrm{e}^{ i \epsilon ^N F_n} \cdots \textrm{e}^{i \epsilon F_1}\) for suitable self-adjoint matrices \(F_1, \ldots , F_N\) and diagonal matrices \(D_1, \ldots , D_N\). \(\square \)
Appendix F: Some Black-Box Lemma
We collect some known 1D resolvent estimates as black boxes. All will be used in Sect. 4, when reducing the resolvent estimate to the one-dimensional setting. The first estimate is now well-known as the geometric control estimate:
Lemma F.1
[23] Let \(I\subset {\mathbb {T}}\) be a non-empty open set. Then there exists \(C=C_I>0\), such that for any \(v\in L^2({\mathbb {T}})\), \(f_1\in L^2({\mathbb {T}}), f_2\in H^{-1}({\mathbb {T}})\), \(\lambda \ge 1\), if
we have
This result can be deduced from the one-dimensional uniform stabilization for the wave equation in [23]. The passage from the uniform stabilization to the resolvent estimate can be also found in Proposition 4.2 and “Appendix A” of [15], or the proof of Proposition 1.4 of [21].
The second estimate follows from the sharp resolvent estimate for the damped-wave operator on \({\mathbb {T}}^2\) with rectangular-shaped damping:
Lemma F.2
[24, Formula (6)]. There exists \(h_0>0\), \(C>0\) such that for all \(0<h<h_0\), \(E\in {\mathbb {R}}\), for any solution v of
we have
The third estimate is the almost sharp resolvent estimate for the damped-wave operator on \({\mathbb {T}}^2\), proved in [2] and (essentially) revisited in [45]:
Lemma F.3
[2, Theorem 2.6]. There exists \(h_0>0\), \(C>0\) and \(\delta _0=\delta _0(\sigma )\) such that for all \(0<h<h_0\), for any solution v of
we have
The last estimate is a special case of the sharp resolvent estimate for the damped-wave operator on \({\mathbb {T}}^2\) for the narrowly undamped situation:
Lemma F.4
[34, Theorem 1.8]. There exists \(h_0>0\), \(C>0\), such that for all \(0<h<h_0\), for any solution v of
we have
We used also intensively a commutator estimate for Lipschitz functions:
Lemma F.5
Assume that \(\kappa \in W^{1,\infty }({\mathbb {R}}^d)\) and \(a\in S^0({\mathbb {R}}^{2d})\), then
The proof of this Lemma is standard and can be found, for example, in Corollary (A.2) of [?]. The proof there applies to Weyl quantization as well. In various places of this article, we apply Lemma F.5 to deduce that \([b^{1/2},{\text {Op}}_h^{\textrm{w}}(a)]=O_{{\mathcal {L}}(L^2)}(h)\) and \([b^{1/2},[b^{1/2},{\text {Op}}_h^{\textrm{w}}(a)]]=O_{{\mathcal {L}}(L^2)}(h^2)\), thanks to the hypothesis (1.5).
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Arnaiz, V., Sun, C. Sharp Resolvent Estimate for the Damped-Wave Baouendi–Grushin Operator and Applications. Commun. Math. Phys. 400, 541–637 (2023). https://doi.org/10.1007/s00220-022-04606-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00220-022-04606-4