The dataset were generated using computer random number generation to sample two sets of random points (X_1, ..., X_n) and (Y_1, ..., Y_m), and then the two-sample coverage threshold R_{n,m,k} was calculated for these random points by measuring the Euclidean distances between pairs |X_i - Y_j|.

This was implemented by the code available at https://github.com/frankiehiggs/CovXY (commit 4c4c086).