Added covercorr(), a unified front-end that
dispatches to the appropriate coverage correlation routine based on its
input. It accepts a pair of variables (x, y),
a list of variables, or a matrix/data frame/data table (each column
treated as a variable), and chooses random or deterministic reference
points via the reference argument.
Added coverage_correlation_grid(), a variant of
coverage_correlation() that accepts user-supplied reference
points (u, v) instead of generating them
randomly. When both inputs are one-dimensional and no points are
supplied, it defaults to the deterministic uniform grid
{1/n, ..., 1}, making the rank transformation
reproducible.
Added coverage_correlation_K(), which generalises
the coefficient from a pair of variables to K mutually
compared variables supplied as a list.
Added coverage_correlation_K_grid(), the
K-variable counterpart of
coverage_correlation_grid(), accepting an optional
grid list of reference points and defaulting to the uniform
grid for one-dimensional inputs.
All coverage correlation functions now return an object of class
"covercorr" with print() and
summary() methods for readable output.
The fixed-grid pairwise p-value now uses the correct null centering for the deterministic-grid case, including the second-order term in the expansion of the null mean.
Explicitly supplied M (Monte Carlo sample size) is
now coerced to an integer and validated, so passing a plain numeric
value no longer triggers a low-level error.
NA handling now preserves matrix structure
throughout (using drop = FALSE), and rows are dropped
consistently across all inputs and reference grids.
K-variable (K > 2) fixed-grid case. In that
case coverage_correlation_K_grid() returns
pval = NA (with pval_available = FALSE) and
emits an informative message; the statistic itself is still computed.
The K = 2 fixed-grid case returns a valid p-value
consistent with coverage_correlation_grid().coverage_correlation() for the
coverage correlation coefficient between two random variables or
vectors, with exact and Monte Carlo computation methods and optional
visualisation.