fiedler_ordering.RdBuilds a kNN graph on rows of X, computes a 1D ordering coordinate using
the Fiedler direction of a normalized graph operator, and returns an ordering
t scaled to \([0,1]\).
Numeric matrix \(n \times D\) (rows are observations).
Number of nearest neighbors.
Similarity type: "rbf", "inv", or "binary".
Bandwidth for weight="rbf"; if NULL, uses median kNN distance.
Component handling: "giant" keeps the largest connected component;
"all" uses all nodes (may be unstable if disconnected).
If TRUE (default), return t of length \(n\)
with NA for nodes excluded by keep="giant".
If FALSE, return t only for the kept nodes.
A list with t, keep_idx, and n_components.