Semi-automatic fitting of covariance functions in 1D and 2D

Giovanna Venuti (mail)

Mirko Reguzzoni ( mail)

Metodologie e architetture software avanzate

Covariance-functions, Least-squares, Stochastic-modeling

The goal is to write a graphical user interface that supports the covariance modelling in 1D and 2D (in both planar and spherical domain) for stochastic interpolation methods. The software has to read the available data, compute the empirical covariance function, show a set of possible positive definite models and, for each of them, allow the manual setting of their parameters. When changing these parameters the corresponding covariance function has to be visualized and, when the user is satisfied by the manual fitting, an automatic least-squares adjustment has to be performed.