Convergence diagnostics for a fitted Rceattle model — convergence

Runs the post-fit convergence battery and returns a single structured object. Each check yields a record with a common schema (id, tier, severity, message, data); severity is one of "OK", "NOTE", "WARN", "FAIL". The object's status is the worst severity present.

Usage

convergence_diagnostics(object, ...)

Arguments

object: An object of class "Rceattle" returned by fit_mod().
...: Currently unused.

Value

An object of class "Rceattle_convergence": a list with status (overall worst severity) and checks (named list of records).

Details

fit_mod() runs this automatically and attaches the result as fit$convergence; call convergence_diagnostics() directly to re-run it on any fit. Checks cover the optimizer gradient, Hessian positive-definiteness and conditioning, parameters on bounds, phasing, and parameter estimability.