Software

sFO-MLA

Implementations of the stochastic first-order Mirror Langevin Algorithm and its warm-started two-loop variant for constrained sampling, including applications to Bayesian mixture weights and Poisson graphical models.

Code · Paper

CR-MRF for Ordinal Preference Data

Code for fitting covariate-dependent consecutive-ratio Markov random fields to multivariate ordinal preference data, with comparisons to pseudo-likelihood and Plackett–Luce approaches.

Code · Paper

Exact and Approximate MCMC for Doubly-Intractable Graphical Models

Code accompanying work on exact and approximate MCMC methods for probabilistic graphical models with intractable normalizing constants, leveraging the corresponding independence model to construct efficient samplers.

Code · Paper · arXiv

ExponentialGM

R code for likelihood-based inference in fully and partially observed exponential-family graphical models with intractable normalizing constants.

Code · Paper

EBayes-Integration

R functions for empirical Bayes data integration in multi-response regression problems.

Code · Paper

OCal

R code for orthogonal calibration via posterior projections, including applications to computer model calibration and the Schwarzschild model.

Code · Paper

bigMVP

Software for scalable two-stage Bayesian inference in high-dimensional multivariate probit models for multivariate binary responses.

Code · Paper

NN-DM

R code for nearest-neighbor Dirichlet mixtures, a scalable Bayesian nonparametric model for multivariate density estimation.

Code · Paper

Fractional Probit Model

R code for Bayesian semiparametric long-memory modeling of discretized event data.

Code · Paper

Simultaneous Dimension Reduction and Variable Selection

MATLAB code for Bayesian sparse multiple regression with simultaneous rank reduction and variable selection.

Code · Paper

Fast Sampling of Gaussian Posteriors

MATLAB implementation of a fast exact sampler for Gaussian posterior distributions arising in high-dimensional regression with Gaussian scale-mixture priors.

Code · Paper

horseshoe

R package for high-dimensional regression and normal-means estimation using horseshoe shrinkage priors, including the fast Gaussian posterior sampling strategy above.

CRAN · Paper

Truncated Normal CDF

Rcpp implementation for computing probabilities of truncated multivariate Gaussian distributions using importance sampling.

Code

Additional research code and reproducibility material are available on my GitHub profile. Bug reports and suggestions are welcome.