Citing THOTH’s Methods
We provide a BibTex file for the necessary references. You are also welcome, as you please, to reference this site, http://thoth-python.org.
bs: Bootstrap Estimators / Error Characterization
The bootstrap estimators were first described and characterized by DeDeo, Hawkins, Klingenstein and Hitchcock in 2013.
In addition, the methods for characterizing arbitrary estimators for bias, reliability, and consistency under coarse graining were presented in that paper.
nsb: Bayesian Estimators (NSB)
The entropy_nsb and mi_nsb functions use a mixture of Dirichlet distributions that is approximately uniform in entropy. This method, often called NSB, was proposed by Nemenman, Shafee and Bialek in 2001. As of writing, it is requested on the NSB site that you reference this work for completeness.
ww: Bayesian Estimators (WW)
The entropy_ww and mi_ww functions use a Dirichlet distribution with beta equal to unity. This estimator was first described by Wolpert and Wolf in 1995. Some minor corrections were reported in 2013.