A Bernstein copula is a nonparametric copula , which can be described by a given grid resolution and an associated discrete distribution skeleton. The (minimal) discrete skeleton can be extracted in an unambiguous manner directly from a multivariate data set , if the grid constants is given by the sample size. A major problem arises when the grid constants is not equal to the sample size (either greater or less ) . The known algorithms for this case to estimate a suitable discrete distribution skeleton are memory intensive and therefore particularly unsuitable for higher dimensions .
The aim of the project is to determine, for any fixed grid constants, an appropriate discrete skeletons from multivariate data so that the memory computational effort is minimal.
Attachment:- BCopulas.pdf