Projects

If you’d like to get involved in one of these projects or have another project in mind, please let me know. My research is in computational statistics algorithms and applied statistics methodology. I work on development for Stan’s probabilistic programming language and automatic differentiation library.

Algorithms

Parallel Hamiltonian Monte Carlo

HMC sampling is embarrassingly parallelizable after stationarity; the sequential bottlenecks are finding the first reasonable draw and adapting to posterior covariance.

Integrators for stiff Hamiltonians

The potential energy field induced by the negative log posterior can have varying curvature beyond the ability of a Euclidean metric to compensate, inducing stiffness in the leapfrog integrator and causing HMC to mix poorly.

Stan

Language design

The heart of Stan is its probabilistic programming language. My main development project is enhancing the language.

Automatic differentiation library

Testing

Documentation

Interfaces

Books

Applied Statistics