R/MakePD.R
check_transform_positivedefinit.RdGiven a list of variance-covariance matrices passed by the parameter pl_mat,
each of the matrices is checked whether it is positive definite based on the smallest
eigenvalue of the matrix. Non-positive definite matrices are bent using a method
that is determined based on the parameters given. If the function is called with
just the list of matrices, then the Schaeffer method is used using the function
makePD2(). Specifing a ratio between largest and smallest eigenvalue using
the parameter pn_ratio causes the usage of the ratio method. When specifying
a minimal eigenvalue using parameter pn_eps, the method of Jorjani et al.
2003 is used. This method exists in a weighted and an unweighted version. Which
of the two versions is used is determined by the parameter pmat_weight. When a
weight matrix is given via pmat_weight the weighted version is used, otherwise
the unweighted version is used.
check_transform_positivedefinit( pl_mat, pn_ratio = NULL, pn_eps = NULL, pmat_weight = NULL, pn_digits = NULL, pb_log = FALSE, plogger = NULL )
| pl_mat | list of input matrices which are bent if necessary |
|---|---|
| pn_ratio | specified ratio between largest and smallest eigenvalue |
| pn_eps | smallest eigenvalue in weighted bending |
| pmat_weight | weight matrix for weighted bending |
| pn_digits | number of digits to be rounded to |
| pb_log | indicator flag for logging |
| plogger | log4r object to log to |