Abstract
We continue the study of stability of solving the interior problem of tomography. The starting point is the Gelfand–Graev formula, which converts the tomographic data into the finite Hilbert transform (FHT) of an unknown function f along a collection of lines. Pick one such line, call it the x-axis, and assume that the function to be reconstructed depends on a one-dimensional argument by restricting f to the x-axis. Let \(I_1\) be the interval where f is supported, and \(I_2\) be the interval where the Hilbert transform of f can be computed using the Gelfand–Graev formula. The equation to be solved is \(\left. {\mathcal {H}}_1 f=g\right| _{I_2}\), where \({\mathcal {H}}_1\) is the FHT that integrates over \(I_1\) and gives the result on \(I_2\), i.e. \({\mathcal {H}}_1: L^2(I_1)\rightarrow L^2(I_2)\). In the case of complete data, \(I_1\subset I_2\), and the classical FHT inversion formula reconstructs f in a stable fashion. In the case of interior problem (i.e., when the tomographic data are truncated), \(I_1\) is no longer a subset of \(I_2\), and the inversion problems becomes severely unstable. By using a differential operator L that commutes with \({\mathcal {H}}_1\), one can obtain the singular value decomposition of \({\mathcal {H}}_1\). Then the rate of decay of singular values of \({\mathcal {H}}_1\) is the measure of instability of finding f. Depending on the available tomographic data, different relative positions of the intervals \(I_{1,2}\) are possible. The cases when \(I_1\) and \(I_2\) are at a positive distance from each other or when they overlap have been investigated already. It was shown that in both cases the spectrum of the operator \({\mathcal {H}}_1^*{\mathcal {H}}_1\) is discrete, and the asymptotics of its eigenvalues \(\sigma _n\) as \(n\rightarrow \infty \) has been obtained. In this paper we consider the case when the intervals \(I_1=(a_1,0)\) and \(I_2=(0,a_2)\) are adjacent. Here \(a_1 < 0 < a_2\). Using recent developments in the Titchmarsh–Weyl theory, we show that the operator L corresponding to two touching intervals has only continuous spectrum and obtain two isometric transformations \(U_1\), \(U_2\), such that \(U_2{\mathcal {H}}_1 U_1^*\) is the multiplication operator with the function \(\sigma (\lambda )\), \(\lambda \ge (a_1^2+a_2^2)/8\). Here \(\lambda \) is the spectral parameter. Then we show that \(\sigma (\lambda )\rightarrow 0\) as \(\lambda \rightarrow \infty \) exponentially fast. This implies that the problem of finding f is severely ill-posed. We also obtain the leading asymptotic behavior of the kernels involved in the integral operators \(U_1\), \(U_2\) as \(\lambda \rightarrow \infty \). When the intervals are symmetric, i.e. \(-a_1=a_2\), the operators \(U_1\), \(U_2\) are obtained explicitly in terms of hypergeometric functions.
Similar content being viewed by others
References
Al-Aifari, R., Defrise, M., Katsevich, A.: Asymptotic analysis of the SVD for the truncated Hilbert transform with overlap. SIAM J. Math. Anal. 47, 797–824 (2015)
Al-Aifari, R., Katsevich, A.: Spectral analysis of the truncated Hilbert transform with overlap. SIAM J. Math. Anal. 46, 192–213 (2014)
Abramowitz, M., Stegun, I.: Handbook of Mathematical Functions. Dover, New York (1970)
Bennewitz, C., Everitt, W.N.: The Titchmarsh–Weyl eigenfunction expansion theorem for Sturm–Liouville differential equations. In: Amrein, W.O., Hinz, A.M., Pearson, D.B. (eds.) Sturm–Liouville Theory: Past and Present, pp. 137–171. Birkhauser, Basel (2005)
Bertola, M., Katsevich, A., Tovbis, A.: Singular value decomposition of a finite Hilbert transform defined on several intervals and the interior problem of tomography: the Riemann–Hilbert problem approach. Comm. Pure Appl. Math. (2014). doi:10.1002/cpa.21547. See also arXiv:1402.0216v1
Epstein, C.L., Schotland, J.: The bad truth about Laplace’s transform. SIAM Rev. 50, 504–520 (2008)
Fulton, C.: Titchmarsh–Weyl \(m\)-functions for second-order Sturm–Liouville problems with two singular endpoints. Mathematische Nachrichten 81, 1418–1475 (2008)
Gelfand, I.M., Graev, M.I.: Crofton function and inversion formulas in real integral geometry. Funct. Anal. Appl. 25, 1–5 (1991)
Gradshteyn, I.S., Ryzhik, I.M.: Table of Integrals, Series, and Products, 5th edn. Academic Press, Boston (1994)
Kamke, E.: Handbook of Ordinary Differential Equations. Chelsea, New York (1971)
Katsevich, A.: Singular value decomposition for the truncated Hilbert transform. Inverse Probl. 26, article ID 115011, 12 (2010)
Katsevich, A.: Singular value decomposition for the truncated Hilbert transform: part II. Inverse Probl. 27. article ID 075006, 7 (2011)
Katsevich, A., Tovbis, A.: Finite hilbert transform with incomplete data: null-space and singular values. Inverse Probl. 28(10), Article ID 105006, 28 (2012)
Rosenblum, M.: On the Hilbert matrix, II. Proc Am Math Soc 9, 581–585 (1958)
Acknowledgments
AK would like to thank Professor John Schotland, a discussion with whom at the conference “Mathematical Methods and Algorithms in Tomography” held at the Mathematisches Forschungsinsitut, Oberwolfach, Germany, in August 2014 gave the initial stimulus to this work. The work of A. Katsevich was supported in part by NSF grant DMS-1211164. The work of A. Tovbis was supported in part by NSF grant DMS-1211164.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Eric Todd Quinto.
Rights and permissions
About this article
Cite this article
Katsevich, A., Tovbis, A. Diagonalization of the Finite Hilbert Transform on Two Adjacent Intervals. J Fourier Anal Appl 22, 1356–1380 (2016). https://doi.org/10.1007/s00041-016-9458-x
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00041-016-9458-x