Astronomy and Astrophysics, volume 563A, 136-136 (2014/3-1)
UVMULTIFIT: A versatile tool for fitting astronomical radio interferometric data.
MARTI-VIDAL I., VLEMMINGS W.H.T., MULLER S. and CASEY S.
Abstract (from CDS):
The analysis of astronomical interferometric data is often performed on the images obtained after deconvolving the interferometer's point spread function. This strategy can be understood (especially for cases of sparse arrays) as fitting models to models, since the deconvolved images are already non-unique model representations of the actual data (i.e., the visibilities). Indeed, the interferometric images may be affected by visibility gridding, weighting schemes (e.g., natural vs. uniform), and the particulars of the (non-linear) deconvolution algorithms. Fitting models to the direct interferometric observables (i.e., the visibilities) is preferable in the cases of simple (analytical) sky intensity distributions. We present UVMULTIFIT, a versatile library for fitting visibility data, implemented in a Python-based framework. Our software is currently based on the CASA package, but can be easily adapted to other analysis packages, provided they have a Python API. The user can simultaneously fit an indefinite number of source components to the data, each of which depend on any algebraic combination of fitting parameters. Fits to individual spectral-line channels or simultaneous fits to all frequency channels are allowed. We have tested the software with synthetic data and with real observations. In some cases (e.g., sources with sizes smaller than the diffraction limit of the interferometer), the results from the fit to the visibilities (e.g., spectra of close by sources) are far superior to the output obtained from the mere analysis of the deconvolved images. UVMULTIFIT is a powerful improvement of existing tasks to extract the maximum amount of information from visibility data, especially in cases close to the sensitivity/resolution limits of interferometric observations.
techniques: interferometric - methods: data analysis