mined:Minimum Energy Designs
This is a method (MinED) for mining probability distributions using deterministic sampling which is proposed by
Joseph, Wang, Gu, Lv, and Tuo (2019)
<DOI:10.1080/00401706.2018.1552203>. The MinED samples can be
used for approximating the target distribution. They can be
generated from a density function that is known only up to a
proportionality constant and thus, it might find applications
in Bayesian computation. Moreover, the MinED samples are
generated with much fewer evaluations of the density function
compared to random sampling-based methods such as MCMC and
therefore, this method will be especially useful when the
unnormalized posterior is expensive or time consuming to
evaluate. This research is supported by a U.S. National Science
Foundation grant DMS-1712642.
