The Randomized Midpoint Method for Log-Concave Sampling

Neural Information Processing Systems 

Sampling from log-concave distributions is a well researched problem that has many applications in statistics and machine learning.