|Abstract || Today, I’ll briefly introduce the sparse modeling and dynamical imaging technique which I learned at MIT Haystack Observatory for 3 months through the UST overseas research grant.|
The very long baseline interferometry (VLBI) has provided an unique view to the (sub)milli-arcsecond (mas) scale angular resolution, especially for the very close region to the vicinity of supermassive black hole (SMBH) at the center of active galactic nucleus (AGN). This based on an accurate measurement of the visibilities and fourier-transform it to obtain the final image. However, the u-v trajectory has essential “holes”, not fully filled (i.e., imperfect sampling), so there are infinite solutions (i.e., images). Therefore further a-priori assumptions/constraints are necessary to obtain the most reasonable solution.
To achieve it, lots of algorithms has been suggested and successfully applied (e.g., CLEAN, Maximum Entropy Method). Sparse Modeling is one of the imaging algorithms based on the statistical minimization including the regularizer terms (e.g., sparsity, gradient). It has benefit not only for a great image reconstruction with super-resolution but also its expandability to include more regularizer terms which constrain the scattering and time-variable effects.
One of the fundamental assumptions of the VLBI imaging is that the target source is not variable during an observation (typically, a few hours). However, this will not be a realistic especially for Sgr A*, the closest SMBH from us, which the orbital period of its accretion flow has been thought around 27 minutes. Dynamical imaging is new VLBI imaging technique to construct a movie, not a single two-dimensional image, so that it can follow the rapid variation of the source. In addition, it can be applied to multi-epoch images which can provide smooth, continuous motion of the source over years, espeically the AGN jets.