Compute compact, label-free quality diagnostics for multiset alignment results. Supports mma_align_multiple in both reference and consensus modes.
Value
A list with fields: - mode: "reference" or "consensus" - per_domain: data.frame with per-domain metrics (n, mse, mse_soft, matched, coverage, mean_confidence, converged, iterations) - global: aggregate metrics (mse_weighted, mse_soft_weighted, K, ref_idx, mean_confidence, all_converged)
Details
Metrics include per-domain mean squared alignment error under final assignments (if available) and a soft counterpart using posteriors. For reference mode, errors are measured to the reference domain; for consensus mode, errors are measured to the consensus centers.
Examples
# \donttest{
# Simulate three domains with shared structure
set.seed(42)
X1 <- matrix(rnorm(100), 50, 2)
X2 <- matrix(rnorm(100), 50, 2)
X3 <- matrix(rnorm(100), 50, 2)
# Create hyperdesign object
hd <- as_hyperdesign(list(X1, X2, X3))
# Perform alignment (requires mma_align_multiple)
# result <- mma_align_multiple(hd, ncomp = 2)
# quality <- alignment_quality(result)
# }