Geometrical optics for the fractional Helmholtz equation and applications to inverse problems
Giovanni Covi, Maarten de Hoop, Mikko Salo · 2024
Read Geometrical optics for the fractional Helmholtz equation and applications to inverse problems by Giovanni Covi, Maarten de Hoop, Mikko Salo (2024) 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-2412-14698.txt.
Introduction
Giovanni Covi, Maarten de Hoop, and Mikko Salo build geometrical optics for the fractional Helmholtz equation, a non-local wave model. They use it on inverse problems. They build a parametrix, an operator that gives near solutions, from geometrical optics solutions. The eikonal equation that goes with it turns out the same as in the classic case. Non-local traits show only in the zeroth-order term of the first transport equation.
The near solutions act in distinct ways across two regimes of that exponent. One is like the classic case. In the other, the potential acts as a strong shift that changes how singularities spread. As a use, the team studies how to recover the potential from Cauchy data when the refraction index is fixed and simple. They prove Holder stability for this inverse problem.
Catalog record
| CREATOR | Giovanni Covi, Maarten de Hoop, Mikko Salo |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 91 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-14698
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Geometrical optics for the fractional Helmholtz equation and applications to inverse problems Giovanni Covi, Maarten de Hoop, Mikko Salo Abstract . In this paper we construct a parametrix for the fractional Helmholtz equation ( ( − Δ ) s − τ 2 s r ( x ) 2 s + q ( x ) ) u = 0 ((-\Delta)^{s}-\tau^{2s}r(x)^{2s}+q(x))u=0 making use of geometrical optics solutions. We show that the associated eikonal equation is the same as in the classical case, while in the first transport equation the effect of nonlocality is only visible in the zero-th order term, which depends on s s . Moreover, we show that the approximate geometrical optics solutions present different behaviors in the regimes s ∈ ( 0 , 1 2 ) s\in(0,\frac{1}{2}) and s ∈ [ 1 2 , 1 ) s\in[\frac{1}{2},1) . While the latter case is quite similar to the classical one, which corresponds to s = 1 s=1 , in the former case we find that the potential is a strong perturbation, which changes the propagation of singularities. As an application, we study the inverse problem consisting in recovering the potential q q from Cauchy data when the refraction index r r is fixed and simple. Using our parametrix based on the construction of approximate geometrical optics solutions, we prove that Hölder stability holds for this problem. This is a substantial improvement over the state of the art for fractional wave equations, for which the usual Runge approximation argument can provide only logarithmic stability. Besides its mathematical novelty, this study is motivated by envisioned applications in nonlocal elasticity models emerging from the geophysical sciences. 1. Introduction In this article we study the fractional Helmholtz operator L=((-\Delta)^{s}-\tau^{2s}r(x)^{2s}+q(x)) and the fractional Helmholtz equation Lu=0\text{ in $\Omega$}. As we will explain below, this is a scalar model for equations appearing in nonlocal elasticity. We assume that \Omega\subset{\mathbb{R}}^{n} is a bounded domain with smooth boundary, r\in C^{\infty}({\mathbb{R}}^{n}) is a positive function (index of refraction), q\in C^{\infty}({\mathbb{R}}^{n}) is a potential, and \tau>0 is a (large) frequency. The operator (-\Delta)^{s} is the fractional Laplacian, defined via the Fourier multiplier |\xi|^{2s} where 0<s<1 . We consider functions u\in H^{s}({\mathbb{R}}^{n}) that solve Lu=0 in \Omega in the sense of distributions. In the classical case where s=1 , the above equation is the standard Helmholtz equation. If the index of refraction r(x) is sufficiently nice (e.g. the Riemannian metric g_{jk}(x)=r(x)^{2}\delta_{jk} in \overline{\Omega} is simple, meaning that it is simply connected with strictly convex boundary and no conjugate points [ 41 ] ), then the equation (-\Delta-\tau^{2}r^{2}+q)u=0 admits geometrical optics solutions that concentrate along geodesics of the metric g when \tau\to\infty . These solutions have the form u=e^{i\tau\varphi}(a_{0}+\tau^{-1}a_{1}+\tau^{-2}a_{2}+\ldots) where the phase function \varphi\in C^{\infty}(\overline{\Omega}) is a real valued solution of the eikonal equation |\nabla\varphi|=r, and the principal amplitude a_{0}\in C^{\infty}(\overline{\Omega}) solves the transport equation 2\nabla\varphi\cdot\nabla a_{0}+(\Delta\varphi)a_{0}=0. Moreover, if we write h=1/\tau and let P=h^{2}r^{-2}L=-h^{2}r^{-2}\Delta-1+h^{2}r^{-2}q , then semiclassical singularities of solutions of Pu=0 propagate along the bicharacteristics of the semiclassical operator P [ 57 , Section 12.3] . The spatial projections of these bicharacteristics are the geodesics of the metric g . We wish to obtain analogous results in the nonlocal case where 0<s<1 . Our first result shows that for 1/2\leq s<1 the fractional Helmholtz equation, even though it is nonlocal, admits geometrical optics solutions that are very similar to the classical case s=1 . Theorem 1.1 (Case 1/2\leq s<1 ) . Suppose that \varphi\in C^{\infty}({\mathbb{R}}^{n}) solves the eikonal equation |\nabla\varphi|=r\text{ in $\Omega$}, and a_{0}\in C^{\infty}_{c}({\mathbb{R}}^{n}) solves the transport equation \left.\begin{array}[]{rl}2\nabla\varphi\cdot\nabla a_{0}+b_{s}a_{0}=0\text{ in $\Omega$},&\text{if $1/2<s<1$,}\\[5.0pt] 2\nabla\varphi\cdot\nabla a_{0}+b_{s}a_{0}=q\text{ in $\Omega$},&\text{if $s=1/2$,}\end{array}\right. where b_{s}:=\Delta\varphi+(2s-2)\frac{\varphi^{\prime\prime}\nabla\varphi\cdot\nabla\varphi}{\left\lvert\nabla\varphi\right\rvert^{2}} . Then for any N\in\mathbb{N} there is a solution of Lu=0 in \Omega having the form u=e^{i\tau\varphi}(a_{0}+\tau^{-\alpha_{1}}a_{1}+\ldots+\tau^{-\alpha_{N}}a_{N})+R_{N} where a_{1},\ldots,a_{N}\in C^{\infty}_{c}({\mathbb{R}}^{n}) are independent of \tau , and R_{N}\in H^{s}({\mathbb{R}}^{n}) satisfies \left\lVert R_{N}\right\rVert_{H^{s}}=O(\tau^{-C_{s}N}) for a fixed constant C_{s}>0 depending only on s . Above, (\alpha_{j})_{j=1}^{\infty}\subset{\mathbb{R}}_{+} is a strictly increasing sequence depending on s such that \alpha_{j}\to\infty as j\to\infty . We make the following remarks: • The eikonal equation for 1/2\leq s<1 is exactly the same as for s=1 , which indicates that singularities propagate along geodesics. • The first transport equation for 1/2<s<1 is almost the same as for s=1 . The effect of nonlocality is only visible in the zeroth order term b_{s} that depends on s . Thus a_{0} will contain an exponential s -dependent attenuation factor. • For s=1/2 the first transport equation also contains the potential q , which can be considered as part of the subprincipal symbol (the principal part is of order 1 and q is of order 0 ). This is analogous to the case of first order perturbations of the Laplacian. • The functions \varphi and a_{j} need to be defined on {\mathbb{R}}^{n} to have u\in H^{s}({\mathbb{R}}^{n}) . Their values outside \overline{\Omega} will affect R_{N} , but not the asymptotics in \tau . We also note that for 1/2\leq s<1 , one could formally apply the semiclassical propagation of singularities theorem in [ 57 , Section 12.3] to conclude that singularities propagate along geodesics. This is not quite rigorous since the symbol of h^{2s}(-\Delta)^{s} is not smooth at \xi=0 . Our proof of Theorem 1.1 is based on stationary phase expansions that deal with this nonsmoothness. Next we state a result for 0<s<1/2 . In this case the potential q is a strong perturbation (the principal part has order 2s<1 and q is of order 0 ), which changes the propagation of singularities. Similar phenomena appear in long range scattering theory [ 30 , Volume 4] . In the geometrical optics construction this will be visible in the presence of lower order phase functions that give rise to exponential factors oscillating at different frequencies. Theorem 1.2 (Case 0<s<1/2 ) . Suppose that \varphi_{0}\in C^{\infty}({\mathbb{R}}^{n}) solves the eikonal equation |\nabla\varphi_{0}|=r\text{ in $\Omega$}, and a_{0}\in C^{\infty}_{c}({\mathbb{R}}^{n}) solves the transport equation 2\nabla\varphi\cdot\nabla a_{0}+b_{s}a_{0}=0\text{ in $\Omega$} where b_{s}:=\Delta\varphi+(2s-2)\frac{\varphi^{\prime\prime}\nabla\varphi\cdot\nabla\varphi}{\left\lvert\nabla\varphi\right\rvert^{2}} . Then for any N\in\mathbb{N} there exist M\in\mathbb{N} and a solution of Lu=0 in \Omega having the form u=e^{i\tau(\varphi_{0}+\tau^{-2s}\varphi_{1}+\ldots+\tau^{-2Ms}\varphi_{M})}(a_{0}+\tau^{-\alpha_{1}}a_{1}+\ldots+\tau^{-\alpha_{M}}a_{M})+R_{N} where \varphi_{j}\in C^{\infty}({\mathbb{R}}^{n}) and a_{j}\in C^{\infty}_{c}({\mathbb{R}}^{n}) are independent of \tau , and R_{N}\in H^{s}({\mathbb{R}}^{n}) satisfies \left\lVert R_{N}\right\rVert_{H^{s}}=O(\tau^{-C_{s}N}) for a fixed constant C_{s}>0 depending only on s . Note that in both Theorem 1.1 and 1.2 , if x_{0}\notin\overline{\Omega} and if (\rho,\theta) are Riemannian normal coordinates for the metric g_{jk}=r(x)^{2}\delta_{jk} centered at x_{0} , then choosing \varphi(x)=\varphi_{0}(x)=\mathrm{dist}_{g}(x,x_{0})=\rho and a_{0}(\rho,\theta)=e^{F(\rho,\theta)}b(\theta) , where F is a certain fixed function depending on r and q , while b is any function supported near \theta_{0} , yields solutions that concentrate near the geodesic \rho\mapsto(\rho,\theta_{0}) . See the proof of Theorem 1.3 for more details. As an application of the geometrical optics construction, we consider the inverse problem consisting in recovering the potential q from Cauchy data in the case of fixed, but unknown, refraction index r . In particular, we have the following Hölder stability result. Theorem 1.3 . Let s\in[\frac{1}{2},1) , \Omega\subseteq{\mathbb{R}}^{n} be open and bounded. Assume that the refraction index r\in C^{\infty}({\mathbb{R}}^{n}) is positive with r\equiv 1 in \mathbb{R}^{n}\setminus\Omega , and that the manifold (\overline{\Omega},g) with g_{jk}=r(x)^{2}\delta_{jk} is simple. Moreover, let k,N\in{\mathbb{N}} , and assume that q_{1},q_{2}\in H^{k}({\mathbb{R}}^{n}) are such that \|q_{1}\|_{H^{k}},\|q_{2}\|_{H^{k}}\leq N , with q_{1}=q_{2} in