Perform the M-Step (Sequential Execution) — perform_m_step

This function is called once per EM iteration from C++. It estimates the parameters for all M states sequentially, properly handling diagnostics collection.

Usage

perform_m_step_r(
  y,
  weights,
  spec,
  model_type,
  diagnostics = NULL,
  iteration = NULL,
  verbose = FALSE,
  dcc_threshold = 0.02,
  dcc_criterion = "bic"
)

Arguments

y: The time series data.
weights: The (T x M) matrix of smoothed probabilities from the E-step.
spec: The full list of model specifications.
model_type: "univariate" or "multivariate".
diagnostics: Diagnostic collector object (ms_diagnostics class).
iteration: Current EM iteration number.
verbose: Logical indicating whether to print progress.
dcc_threshold: dcc_threshold
dcc_criterion: dcc_criterion

Value

A list of length M containing the updated model fits for each state.