Compute pairwise distance matrix

Compute pairwise distance matrix

Usage

compute_distance_matrix(
  left_mat,
  right_mat,
  distance = "euclidean",
  sigma = NULL
)

Arguments

left_mat

Numeric matrix of left units (rows = units, cols = variables).

right_mat

Numeric matrix of right units (rows = units, cols = variables).

distance

Character string specifying distance metric, or a function.

sigma

Optional covariance matrix for Mahalanobis distance. If NULL (default), the pooled within-group covariance is estimated from left_mat and right_mat: \(\Sigma = ((n_L-1)\,S_L + (n_R-1)\,S_R) / (n_L+n_R-2)\). This matches the convention used by optmatch::match_on() for treated/control Mahalanobis matching. If either group has fewer than two rows, falls back to the overall-sample covariance of the two groups stacked together.

Value

Numeric matrix of pairwise distances (n_left x n_right).