KSMMD: Stata module to perform Weighted k-sample distributional test combining Kolmogorov-Smirnov (Kiefer's T) and Maximum Mean Discrepancy (MMD via Random Fourier Features)

ksmmd tests whether k >= 2 groups (named by by()) come from the same distribution of varname, computing two test statistics from a single permutation resample: o a k-sample generalization of the Kolmogorov-Smirnov test (Kiefer's T -- Kiefer 1959), the same statistic kstest (Ariel Linden, SSC s459801) computes, but reimplemented here so that varname is sorted once (not re-sorted on every permutation) -- see Remarks: why reimplement kstest. o a k-sample generalization of the Maximum Mean Discrepancy (MMD -- Gretton et al. 2012) with an RBF kernel, approximated via Random Fourier Features (Rahimi & Recht 2007) so it stays tractable at survey-sized n (~10^5), unlike mmd_2s (Boston College SSC s459820), which only compares 2 groups with an exact O(N^2) kernel.