A novel approach for three-dimensional electromagnetic imaging is presented. This technique is a combination of an iterative multiscaling approach with an inexact-Newton method. The multiscaling procedure allows one to iteratively focus the region of interest on the detected target, whereas the inexact-Newton method provides an efficient regularized solution of the nonlinear electromagnetic inverse scattering problem for each scaling step. The proposed method is validated against numerical data with different configuration settings. A preliminary experimental validation is also reported.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1364/JOSAA.34.001119 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Dynamics simulation with frictional contacts is important for a wide range of applications, from cloth simulation to object manipulation. Recent methods using smoothed lagged friction forces have enabled robust and differentiable simulation of elastodynamics with friction. However, the resulting frictional behavior can be inaccurate and may not converge to analytic solutions.
View Article and Find Full Text PDFGeometry and convergence of natural policy gradient methods.
Inf Geom
June 2023
Max Planck Institute for Mathematics in the Sciences, Inselstraße 22, Leipzig, 04103 Saxony Germany.
We study the convergence of several natural policy gradient (NPG) methods in infinite-horizon discounted Markov decision processes with regular policy parametrizations. For a variety of NPGs and reward functions we show that the trajectories in state-action space are solutions of gradient flows with respect to Hessian geometries, based on which we obtain global convergence guarantees and convergence rates. In particular, we show linear convergence for unregularized and regularized NPG flows with the metrics proposed by Kakade and Morimura and co-authors by observing that these arise from the Hessian geometries of conditional entropy and entropy respectively.
View Article and Find Full Text PDFA multi-scale simulation of retinal physiology.
Math Biosci
September 2023
Department of Mathematics, University of Toronto, Toronto, ON, M5S 2E4, Canada. Electronic address:
We present a detailed physiological model of the (human) retina that includes the biochemistry and electrophysiology of phototransduction, neuronal electrical coupling, and the spherical geometry of the eye. The model is a parabolic-elliptic system of partial differential equations based on the mathematical framework of the bi-domain equations, which we have generalized to account for multiple cell-types. We discretize in space with non-uniform finite differences and step through time with a custom adaptive time-stepper that employs a backward differentiation formula and an inexact Newton method.
View Article and Find Full Text PDFIterative subspace algorithms for finite-temperature solution of Dyson equation.
J Chem Phys
March 2022
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109, USA.
One-particle Green's functions obtained from the self-consistent solution of the Dyson equation can be employed in the evaluation of spectroscopic and thermodynamic properties for both molecules and solids. However, typical acceleration techniques used in the traditional quantum chemistry self-consistent algorithms cannot be easily deployed for the Green's function methods because of a non-convex grand potential functional and a non-idempotent density matrix. Moreover, the optimization problem can become more challenging due to the inclusion of correlation effects, changing chemical potential, and fluctuations of the number of particles.
View Article and Find Full Text PDFDistributed second-order optimization, as an effective strategy for training large-scale machine learning systems, has been widely investigated due to its low communication complexity. However, the existing distributed second-order optimization algorithms, including distributed approximate Newton (DANE), accelerated inexact DANE (AIDE), and statistically preconditioned accelerated gradient (SPAG), are all required to precisely solve an expensive subproblem up to the target precision. Therefore, this causes these algorithms to suffer from high computation costs and this hinders their development.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!