README.md

In jlisic/lpm3: A Fast K-D Tree Implementation of the Local Pivotal Method

This is a "faster" implementation of the Local Pivotal Method for balanced sampling.

Current timings relative to BalancedSampling's lpm2 procedure on an i5-4258U CPU @ 2.40GHz running R 3.2.1 on OS X 10.10.4:

N = 100,000 n = 1,000 x = cbind(1:N, 1:N)

lpm2 (Balanced Sampling): run time = 43.708 elapsed seconds

lpm3: run time = 0.183 elapsed seconds (Note, this includes the time to generate x, but does not include R function call and copying overhead).

jlisic/lpm3 documentation built on May 19, 2019, 12:46 p.m.