High Resolution Inverse Scattering in Two Dimensions Using Recursive Linearization

Abstract

Abstract. We describe a fast, stable algorithm for the solution of the inverse acoustic scattering problem in two dimensions. Given full aperture far field measurements of the scattered field for multiple angles of incidence, we use Chen's method of recursive linearization to reconstruct an unknown sound speed at resolutions of thousands of square wavelengths in a fully nonlinear regime. Despite the fact that the underlying optimization problem is formally ill-posed and non-convex, recursive linearization requires only the solution of a sequence of linear least squares problems at successively higher frequencies. By seeking a suitably band-limited approximation of the sound speed profile, each least squares calculation is well-conditioned and involves the solution of a large number of forward scattering problems, for which we employ a recently developed, spectrally accurate, fast direct solver. For the largest problems considered, involving 19,600 unknowns, approximately one million partial differential equations were solved, requiring approximately two days to compute using a parallel MATLAB implementation on a multi-core workstation.

An Adaptive High Order Direct Solution Technique for Elliptic Boundary Value Problems

“…The numerical results in that paper showed the method did not observe pollution for problems where the support of the deviation from constant coefficient was 100 time the smallest wavelength in size. In [5], the method was utilized to build a inverse scattering solver via the recursive linearization procedure proposed in [9]. The recursive linearization procedure requires solving a sequence of linear least squares problems at successively higher frequencies to reconstruct an unknown sound speed.…”

“…Compute the y-directional Chebychev coefficients B j of the approximate solution on τ . (5) end for (6) Let S y = max j=2,...,nc−1 (|B j (n c − 1)| + |B j (n c ) − B j (n c − 2)|) (7) for j = 2, . .…”

Rapid Solution of the Cryo-EM Reconstruction Problem by Frequency Marching

“…In the language of optimization, this can be viewed as a homotopy method using the maximum spatial frequency (resolution) as the homotopy parameter. The basic intuition underlying our scheme is motivated by the success of recursive linearization in inverse acoustic scattering [1,4,6,7].…”

“…For each image m we first defined a realistic radially-symmetric CTF function C (m) (k) using the standard formulae [28,54] Here χ is called the phase function, θ 0 = 0.002 sets the microscope acceptance angle, w 2 = 0.07 controls the relative inelastic scattering, with w 2 1 + w 2 2 = 1, and the spherical aberration is C s = 2 × 10 7Å . The defocus parameter z was different for each image, chosen uniformly at random in the interval [1,4] where λ = 0.025Å is the free-space electron wavelength (a typical value corresponding to a 200 keV microscope), and for convenience the distance scaling D has been included. In our setting θ is of order 10 −5 times the numerical wavenumber k.…”

Localization of small obstacles from back-scattered data at limited incident angles with full-waveform inversion

“…It is also the principle in Recursive linearization Algorithm [25], see discussions in [43,60,16]. Multifrequencies data are also used in [34,11,17] for location of homogeneities, [7,10,15] for medium reconstruction, [16] for shape reconstruction.…”