A more complex function is the Maximum Wishart Likelihood (ML) discrepancy function. This function may be written

(56)

where Tr( ) denotes the trace operator, |S| the determinant of **S**, the
*unbiased* sample covariance matrix, and *p* is the number of
manifest variables. If **S** has a Wishart distribution (a somewhat
less restrictive assumption than the requirement that the observed variables
follow a multivariate normal distribution), minimizing the ML discrepancy
function produces *Maximum Wishart Likelihood Estimates*, i.e.,

(57)