Dichroic polarization at mid-infrared wavelengths: a Bayesian approach

TL;DR

This paper introduces a Bayesian inference framework for analyzing dichroic polarization in the mid-infrared, enabling more precise estimation of magnetic fields and polarization components from imaging and spectro-polarimetric data.

Contribution

The paper presents a novel Bayesian hierarchical regression method that incorporates dust properties and extinction curves to improve polarization analysis.

Findings

Allows customized absorptive polarization modeling based on dust composition.

Provides more accurate magnetic field strength and geometry estimations.

Recommends specific filter combinations for optimal polarization mechanism disentanglement.

Abstract

A fast and general Bayesian inference framework to infer the physical properties of dichroic polarization using mid-infrared imaging- and spectro-polarimetric observations is presented. The Bayesian approach is based on a hierarchical regression and No-U-Turn Sampler method. This approach simultaneously infers the normalized Stokes parameters to find the full family of solutions that best describe the observations. In comparison with previous methods, the developed Bayesian approach allows the user to introduce a customized absorptive polarization component based on the dust composition, and the appropriate extinction curve of the object. This approach allows the user to obtain more precise estimations of the magnetic field strength and geometry for tomographic studies, and information about the dominant polarization components of the object. Based on this model, imaging-polarimetric observations using two or three filters located in the central 9.5-10.5 $\mu$m, and the edges 8-9 $\mu$m and/or 11-13 $\mu$m, of the wavelength range are recommended to optimally disentangle the polarization mechanisms.