Check EM Monotonicity

Verify that the EM algorithm exhibits (near) monotonic log-likelihood improvement.

Usage

check_em_monotonicity(diagnostics, tolerance = 1e-04)

Arguments

diagnostics

An object of class ms_diagnostics

tolerance

Numeric threshold for acceptable decreases (default: 1e-4)

Value

List with passed (logical), n_violations, violation_iters

Examples

if (FALSE) { # \dontrun{
monotonicity_check <- check_em_monotonicity(diag)
if (!monotonicity_check$passed) {
  cat("Violations at iterations:", monotonicity_check$violation_iters, "\n")
}
} # }