After completing a good observation, doing robust data reduction and finalizing the measurements, it is commonly necessary to parameterize the derived correlations. For example, you have derived the radial profile of the PSF of your image (see Building the extended PSF). You may want to parameterize the radial profile to estimate the slope. Alternatively, you may have found the star formation rate and stellar mass of your sample of galaxies. Now, you want to derive the star formation main sequence as a parametric relation between the two. The fitting functions below can be used for such purposes.
Gnuastro’s least squares fitting features are just wrappers over the least squares fitting methods of the linear and nonlinear least-squares fitting functions of the GNU Scientific Library (GSL). For the low-level details and equations of the methods, please see the GSL documentation. The names have been preserved here in Gnuastro to simplify the connection with GSL and follow the details in the detailed documentation there.
GSL is a very low-level library, designed for maximal portability to many scenarios, and power. Therefore calling GSL’s functions directly for a fast operation requires a good knowledge of the C programming language and many lines of code. As a low-level library, GSL is designed to be the back-end of higher-level programs (like Gnuastro). Through the Statistics program, in Gnuastro we provide a high-level interface to access to GSL’s very powerful least squares fitting engine to read/write from/to standard data formats in astronomy. A fully working example is shown below.
To activate fitting in Statistics, simply give your desired fitting method to the --fit option (for the full list of acceptable methods, see Fitting options). Statistics accepts both 1-dimensional and 2-dimensional inputs which we’ll describe in the sub-sections below.
GNU Astronomy Utilities 0.24 manual, November 2025.