Reconstruction of singularities from full scattering data by new estimates of bilinear Fourier multipliers (Q845754)

From MaRDI portal
Revision as of 05:07, 10 December 2024 by Import241208061232 (talk | contribs) (Normalize DOI.)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Reconstruction of singularities from full scattering data by new estimates of bilinear Fourier multipliers
scientific article

    Statements

    Reconstruction of singularities from full scattering data by new estimates of bilinear Fourier multipliers (English)
    0 references
    29 January 2010
    0 references
    For some complex valued potential \(q(x)\), let \(H=\Delta + q(x)\) be the Schrödinger operator. Given the energy \(k^{2}\) and the incident direction \(\theta\), the scattering solution \(\psi=\psi(x,\theta,k)\) is defined as the solution of the problem \[ \begin{cases} (\Delta+q+k^{2})\psi=0\\ \psi=e^{ikx\cdot\theta}+\psi_{out}, \end{cases} \] where the function \(\psi_{out}\) satisfies the outgoing Sommerfeld radiation condition. This means, for a compactly supported potential \(q\), that it has asymptotics as \(|x|\to\infty\) \[ \psi(x,\theta,k)=e^{ikx\cdot\theta}+C|x|^{\frac{1-n}{2}}k^{\frac{n-3}{2}}e^{ik|x|}A(\theta ', \theta, k)+o(|x|^{\frac{1-n}{2}}), \] where \(\theta'=x/|x|\). Let \(q_{b}\) denote the Born approximation of \(q\), and let \(q_{B}\) be an averaged Born approximation. The following theorem is obtained: Assume that \(q\) is a compactly supported potential, which may be complex valued, such that \(q\in L_{p}(\mathbb R^{n})\cap W^{\alpha,2}(\mathbb R^{n})\), where \(p>n/2\) and \(p\geq 2\), \(\max\{0,\frac{n}{2}-\frac{2n}{n+1}\}\leq\alpha\). Then \(q-q_{B}\in W_{loc}^{\beta,2}\), when \(\beta-\alpha<\epsilon(\alpha)\). The accuracy function \(\epsilon(\alpha)\) is given by: (a) In the interval \(\max\{0,\frac{n}{2}-\frac{2n}{n+1}\}\leq\alpha\leq\frac{n}{2}\) by \[ \epsilon(\alpha)=\begin{cases} \alpha+2-\frac{n}{2}-\frac{\alpha}{n}, &\text{if } \alpha\leq\frac{n}{2}-\frac{n}{2(n-1)} \\ 1, &\text{if } \frac{n}{2}-\frac{n}{2(n-1)}\leq\alpha\leq\frac{n}{2} \end{cases} \] (b) For \(\alpha >\frac{n}{2}\), \(\epsilon(\alpha)\) is obtained from the periodic extension of its values in the interval \((\frac{n}{2}-1,\frac{n}{2}]\). To be precise \[ \epsilon(\alpha)=\max\{\epsilon(\alpha '), \frac{n}{2}-\alpha '\} \] where \(\alpha '=\alpha-1-[\alpha-\frac{n}{2}]\). It means, in particular, that the singularities of a complex valued potential \(q\) can be reconstructed from the linear Born approximation for full scattering data by averaging in the extra variables; besides the accuracy in the reconstruction improves the previously known accuracy obtained from fixed angle or backscattering data. In particular, for \(q\in W^{\alpha,2}\) for \(\alpha\geq 0\), in 2D there is recovered the main singularity of \(q\) with an accuracy of one derivative; in 3D the accuracy is \(\epsilon >\frac{1}{2}\), increasing with \(\alpha\). This gives a mathematical basis for diffraction tomography. The proof is based on some new estimates for multidimensional bilinear Fourier multipliers of independent interest.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers