arXiv subjects

Olivia Tsang

Publications and source records attributed to Olivia Tsang.

4 recordsLinked to original sources

arXiv2025

A Model-Guided Neural Network Method for the Inverse Scattering Problem

Inverse medium scattering is an ill-posed, nonlinear wave-based imaging problem arising in medical imaging, remote sensing, and non-destructive testing. Machine learning (ML) methods offer increased inference speed and flexibility in capturing prior knowledge of imaging targets relative to classical optimization-based approaches; however, they perform poorly in regimes where the scattering behavior is highly nonlinear. A key limitation is that ML methods struggle to incorporate the physics governing the scattering process, which are typically inferred implicitly from the training data or loosely enforced via architectural design. In this paper, we present a method that endows a machine learning framework with explicit knowledge of problem physics, in the form of a differentiable solver representing the forward model. The proposed method progressively refines reconstructions of the scattering potential using measurements at increasing wave frequencies, following a classical strategy to stabilize recovery. Empirically, we find that our method provides high-quality reconstructions at a fraction of the computational or sampling costs of competing approaches.

physics.comp-ph↗

arXiv2025

Multi-Frequency Progressive Refinement for Learned Inverse Scattering

Interpreting scattered acoustic and electromagnetic wave patterns is a computational task that enables remote imaging in a number of important applications, including medical imaging, geophysical exploration, sonar and radar detection, and nondestructive testing of materials. However, accurately and stably recovering an inhomogeneous medium from far-field scattered wave measurements is a computationally difficult problem, due to the nonlinear and non-local nature of the forward scattering process. We design a neural network, called Multi-Frequency Inverse Scattering Network (MFISNet), and a training method to approximate the inverse map from far-field scattered wave measurements at multiple frequencies. We consider three variants of MFISNet, with the strongest performing variant inspired by the recursive linearization method--a commonly used technique for stably inverting scattered wavefield data--that progressively refines the estimate with higher frequency content. MFISNet outperforms past methods in regimes with high-contrast, heterogeneous large objects, and inhomogeneous unknown backgrounds.

physics.comp-ph↗

arXiv2007

A synchrotron self-Compton model with low energy electron cut-off for the blazar S5 0716+714

Rapid inverse Compton cooling sets in when the brightness temperature (T_B) of a self-absorbed synchrotron source with power-law electrons reaches ~10^{12} K. However, T_B inferred from observations of intra-day variable sources (IDV) are well above the "Compton catastrophe" limit. This can be understood if the underlying electron distribution cuts off at low energy. We approximate a low-energy cut-off with monoenergetic electrons. We compute the synchrotron self-Compton (SSC) spectrum of such distribution, and using the IDV source S5~0716+714 as an example, we compare it to the observed SED of S5~0716+714. The hard radio spectrum is well-fitted by this model, and the optical data can be accommodated by a power-law extension to the electron spectrum. We therefore examine the scenario of an injection of electrons that is a double power law in energy with a hard low-energy component that does not contribute to the synchrotron opacity. We show that the double power-law injection model is in good agreement with the observed SED of S5~0716+714. For intrinsic variability, we find that a Doppler factor of D\geq30 can explain the observed SED provided that low-frequency (<32 GHz) emission originates from a larger region than the higher-frequency emission. To fit the entire spectrum, D\geq65 is needed. We find the constraint imposed by induced Compton scattering at high T_B is insignificant in our model. We confirm that electron distribution with a low-energy cut-off can explain the high T_B in compact radio sources. We show that synchrotron spectrum from such distributions naturally accounts for the observed hard radio continuum with a softer optical component, without the need for an inhomogeneous source.

astro-ph↗

arXiv2006

The inverse Compton catastrophe and high brightness temperature radio sources

The occurrence of the inverse Compton catastrophe when the synchrotron brightness temperature exceeds a threshold value, usually estimated to be 10^{12} K, appears to be in contradiction with observation because: (i) the threshold is substantially exceeded by several intra-day variable radio sources, but the inverse Compton emission is not observed, (ii) powerful, extra-galactic radio sources of known angular size do not appear to congregate close to the predicted maximum brightness temperature. We re-examine the parameter space available to synchrotron sources using a population of monoenergetic electrons, in order to see whether the revised threshold temperature is consistent with the data. The electron distribution and the population of each generation of scattered photons are computed using spatially averaged equations. We confirm our previous finding that intrinsic brightness temperatures T_{\rm B}~10^{14} K can occur without catastrophic cooling. We show that substantially higher temperatures cannot be achieved either in transitory solutions or in solutions that balance losses with a powerful acceleration mechanism. Depending on the observing frequency, we find strong cooling can set in at a range of threshold temperatures and the imposition of the additional constraint of equipartition between particle and magnetic field energy is not warranted by the data. Postulating a monoenergetic electron distribution, which approximates one that is truncated below a certain Lorentz factor, γ_{min}, alleviates several theoretical difficulties associated with the inverse Compton catastrophe, including those mentioned above.

astro-ph↗